feat(server): add initAgent conflict_mode overwrite mode with SDK defaults by lan17 · Pull Request #40 · agentcontrol/agent-control

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 · 2026-02-23T05:55:00Z

Summary

add server-side conflict_mode on initAgent (strict default, overwrite opt-in)
in overwrite mode, replace existing steps/evaluators with latest init payload and return overwrite metadata (overwrite_applied, overwrite_changes, warnings)
keep strict mode semantics unchanged
warn (do not block) when overwrite removes evaluators referenced by active controls
update Python SDK init/register flows to send conflict_mode, defaulting SDK init flows to overwrite
pin Python refresh_controls_async() to conflict_mode="strict" to avoid accidental schema removal on refresh
regenerate TypeScript SDK models for new request/response schema and default conflictMode to overwrite in generated outbound schema
add Python and TypeScript tests for conflict-mode defaults/wiring

Story

https://app.shortcut.com/galileo/story/55195/server-make-step-checks-in-initagent-optional

Validation

make server-test
make openapi-spec-check
uv run --package agent-control-server ruff check --config pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py server/tests/test_init_agent_conflict_mode.py
uv run --package agent-control-server mypy --config-file pyproject.toml models/src/agent_control_models/agents.py server/src/agent_control_server/endpoints/agents.py
make sdk-lint
make sdk-typecheck
make sdk-test
make sdk-ts-typecheck
make sdk-ts-test
make sdk-ts-lint
make sdk-ts-generate-check

codecov · 2026-02-23T05:57:54Z

Codecov Report

❌ Patch coverage is 92.00000% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...erver/src/agent_control_server/endpoints/agents.py	91.91%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

…nges

namrataghadi-galileo

Can you show an example that shows this change in action? A comment with an example that you ran in REPL is sufficient. Thanks

lan17 · 2026-02-24T23:12:31Z

Example from a behavioral test (same flow this PR introduces): server/tests/test_init_agent_conflict_mode.py::test_init_agent_overwrite_replaces_steps_and_evaluators.

Scenario:

Initial registration: steps tool-a,tool-b; evaluators eval-a,eval-b
Overwrite registration (conflict_mode="overwrite"): updates tool-a schema/description, adds tool-c, removes tool-b; updates eval-a, adds eval-c, removes eval-b

Expected/observed response shape (asserted in test):

{
"overwrite_applied": true,
"overwrite_changes": {
"steps_added": [{"type": "tool", "name": "tool-c"}],
"steps_updated": [{"type": "tool", "name": "tool-a"}],
"steps_removed": [{"type": "tool", "name": "tool-b"}],
"evaluators_added": ["eval-c"],
"evaluators_updated": ["eval-a"],
"evaluators_removed": ["eval-b"]
}
}

Also covered for refresh behavior: sdks/python/tests/test_init_step_merge.py::test_refresh_controls_uses_strict_conflict_mode, which asserts refresh_controls_async() calls register_agent(..., steps=[], conflict_mode="strict").

namrataghadi-galileo

thank you. this is good!!

## 1.0.0 (2026-03-04) ### ⚠ BREAKING CHANGES * **server:** Feature/56688 fix image bug (#48) * **sdk:** a bug in docker file (#46) * **server:** Feature/56688 fix docker and create bash (#45) * **evaluators:** Evaluator reorganization with new package structure Package Structure: - agent-control-evaluators (v3.0.0): core + regex, list, json, sql - agent-control-evaluator-galileo (v3.0.0): Luna2 evaluator Key Changes: - Entry points for evaluator discovery (agent_control.evaluators) - Dot notation for external evaluators (galileo.luna2 not galileo/luna2) - Dynamic __version__ via importlib.metadata - Server uses evaluators as runtime dep (no longer vendored) - Release workflow publishes both packages to PyPI Bug Fixes: - JSON evaluator: field_constraints/field_patterns in extra-fields allow-list - SQL evaluator: LIMIT/OFFSET bypass fix Migration: - Import: agent_control_evaluator_galileo.luna2 (not agent_control_evaluators.galileo_luna2) - DB: UPDATE controls SET evaluator.name replace('/', '.') * **server:** add time-series stats and split API endpoints (#6) * **evaluators:** rename plugin to evaluator throughout (#81) * **models:** simplify step model and schema (#70) ### Features * Add plugin auto-discovery via Python entry points ([#49](#49)) ([1521182](1521182)) * **docs:** add GitHub badges and CI coverage reporting ([#90](#90)) ([be1fa14](be1fa14)) * **evaluators:** add required_column_values for multi-tenant SQL validation ([#30](#30)) ([532386c](532386c)) * **sdk-ts:** automate semantic-release for npm publishing ([#52](#52)) ([2b43958](2b43958)) * **sdk:** Add PyPI packaging with semantic release ([#52](#52)) ([7c24f7f](7c24f7f)) * **sdk:** Auto-populate init() steps from [@control](https://github.com/control)() decorators ([#23](#23)) ([dc0f2a4](dc0f2a4)) * **sdk:** export ControlScope, ControlMatch, and EvaluatorResult models ([#18](#18)) ([0d49cad](0d49cad)) * **sdk:** Get Agent Controls from SDK Init ([#15](#15)) ([a485f93](a485f93)) * **sdk:** Refresh controls in a background loop ([#43](#43)) ([03f826d](03f826d)) * **sdk:** ship TypeScript SDK with deterministic method naming ([#32](#32)) ([a76e9b0](a76e9b0)) * **server:** add evaluator config store ([#78](#78)) ([cc14aa6](cc14aa6)) * **server:** add initAgent conflict_mode overwrite mode with SDK defaults ([#40](#40)) ([f3ed2b8](f3ed2b8)) * **server:** Add observability system for control execution tracking ([#44](#44)) ([fd0bddc](fd0bddc)) * **server:** add prometheus metrics for endpoints ([#68](#68)) ([775612c](775612c)) * **server:** add time-series stats and split API endpoints ([#6](#6)) ([a0fa597](a0fa597)) * **server:** hard-cut migrate to remove agent UUID ([#44](#44)) ([ee322c9](ee322c9)) * **server:** Optional Policy and many to many relationships ([#41](#41)) ([1a62746](1a62746)) * **ui:** add sql, luna2, json control forms and restructure the code ([#54](#54)) ([c4c1d4a](c4c1d4a)) * **ui:** allow to delete control ([#39](#39)) ([7dc4ca3](7dc4ca3)) * **ui:** Control Store Flow Updated ([#4](#4)) ([dda9f70](dda9f70)) * **ui:** stats dashboard ([#80](#80)) ([4cbb7fe](4cbb7fe)) * **ui:** Steps dropdown rendered based on api return values ([#36](#36)) ([a2aca43](a2aca43)) * **ui:** tests added and some minor ui changes, added error boundaries ([#61](#61)) ([009852b](009852b)) * **ui:** update agent control icon and favicon ([#42](#42)) ([19af8fa](19af8fa)) ### Bug Fixes * **ci:** Add ui scope to PR title validation ([#59](#59)) ([e0fdb52](e0fdb52)) * **ci:** correct galileo contrib path in release build script ([#51](#51)) ([2de6013](2de6013)) * **ci:** Enable pr title on prs ([#56](#56)) ([3d8b5fe](3d8b5fe)) * **ci:** Fix release ([#11](#11)) ([9dd3dd7](9dd3dd7)) * **ci:** Use galileo-automation bot for releases ([#57](#57)) ([bc8eea0](bc8eea0)) * **docs:** Add Example for Evaluator Extension ([#3](#3)) ([c2a70b3](c2a70b3)) * **docs:** add setup script ([#49](#49)) ([7a212c3](7a212c3)) * **docs:** Clean up Protect ([#76](#76)) ([99c16fd](99c16fd)) * **docs:** Fix Examples for LangGraph ([#64](#64)) ([23b30ae](23b30ae)) * **docs:** Improve documentation for open source release ([#47](#47)) ([9018fb3](9018fb3)) * **docs:** Remove old/unused examples ([#66](#66)) ([f417781](f417781)) * **docs:** Update Contributing Guide ([#8](#8)) ([10b34c8](10b34c8)) * **docs:** Update readme ([#37](#37)) ([7531d83](7531d83)) * **docs:** Update README ([#2](#2)) ([379bb15](379bb15)) * **examples:** Control sets cleanup with signed ([#65](#65)) ([af7b5fb](af7b5fb)) * **examples:** Update crew ai example to use evaluator ([#93](#93)) ([1c65084](1c65084)) * **infra:** Add plugins directory to Dockerfile ([#58](#58)) ([171d459](171d459)) * **infra:** install engine/evaluators in server image ([#14](#14)) ([d5ae157](d5ae157)) * **models:** use StrEnum for error enums ([#12](#12)) ([3f41c9f](3f41c9f)) * **sdk-ts:** add conventional commits preset dependency ([#55](#55)) ([540fe9d](540fe9d)) * **sdk-ts:** export npm token for semantic-release npm auth ([#54](#54)) ([1b6b993](1b6b993)) * **sdk:** 54253 add steer action and example ([#38](#38)) ([bf2380a](bf2380a)) * **sdk:** a bug in docker file ([#46](#46)) ([12d1794](12d1794)) * **sdk:** Add step_name as parameter to control ([#25](#25)) ([19ade9d](19ade9d)) * **sdk:** emit observability events for SDK-evaluated controls and fix non_matches propagation ([#24](#24)) ([6a9da69](6a9da69)) * **sdk:** enforce UUID agent IDs ([#9](#9)) ([5ccdbd0](5ccdbd0)) * **sdk:** Fix logging ([#77](#77)) ([b1f078c](b1f078c)) * **sdk:** plugin to evaluator.. agent_protect to agent_control ([#88](#88)) ([fc9b088](fc9b088)) * **server:** enforce public-safe API error responses ([#20](#20)) ([e50d817](e50d817)) * **server:** Feature/56688 fix docker and create bash ([#45](#45)) ([7277e27](7277e27)) * **server:** Feature/56688 fix image bug ([#48](#48)) ([71e6b44](71e6b44)) * **server:** fix alembic migrations ([#47](#47)) ([c19c17c](c19c17c)) * **server:** reject initAgent UUID/name mismatch ([#13](#13)) ([19d61ff](19d61ff)) * tighten evaluation error handling and preserve control data ([52a1ef8](52a1ef8)) * **ui:** Fix UI and clients for simplified step schema ([#75](#75)) ([be2aaf0](be2aaf0)) * **ui:** json validation ([#10](#10)) ([a0cd5af](a0cd5af)) * **ui:** selector subpaths issue ([#34](#34)) ([79cb776](79cb776)) * **ui:** UI feedback fixes ([#27](#27)) ([6004761](6004761)) ### Code Refactoring * **evaluators:** rename plugin to evaluator throughout ([#81](#81)) ([0134682](0134682)) * **evaluators:** split into builtin + extra packages for PyPI ([#5](#5)) ([0e0a78a](0e0a78a)) * **models:** simplify step model and schema ([#70](#70)) ([4c1d637](4c1d637))

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

lan17 added 2 commits February 23, 2026 00:43

feat(server): add initAgent conflict_mode overwrite flow
0f67876

feat(sdk): default initAgent conflict_mode to overwrite
e7fe2d4

lan17 changed the title ~~feat: add initAgent conflict_mode overwrite mode with SDK defaults~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

lan17 changed the title ~~feat(api): add initAgent conflict_mode overwrite mode with SDK defaults~~feat(server): add initAgent conflict_mode overwrite mode with SDK defaultsFeb 23, 2026

refactor(initAgent): move evaluator reference risk into overwrite_cha…
d0cbb5e
…nges

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadmodels/src/agent_control_models/server.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadsdks/python/src/agent_control/__init__.py

namrataghadi-galileo reviewed Feb 24, 2026
View reviewed changes

Comment threadserver/src/agent_control_server/endpoints/agents.py Outdated

lan17 added 2 commits February 24, 2026 17:25

refactor: address initAgent PR review feedback
8235002

chore(sdk-ts): regenerate docs for conflict_mode comments
4668752

namrataghadi-galileo requested changes Feb 24, 2026
View reviewed changes

test: add Given/When/Then comments to initAgent tests
d41b658

namrataghadi-galileo approved these changes Feb 24, 2026
View reviewed changes

lan17 merged commit f3ed2b8 into mainFeb 24, 2026
6 checks passed

lan17 deleted the lev/sc-55195 branch February 24, 2026 23:51

Uh oh!

Conversation

lan17 commented Feb 23, 2026

Summary

Story

Validation

Uh oh!

codecovBot commented Feb 23, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

Uh oh!

namrataghadi-galileo left a comment

Choose a reason for hiding this comment

Uh oh!

lan17 commented Feb 24, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

namrataghadi-galileo left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

lan17 commented Feb 23, 2026

Summary

Story

Validation

Uh oh!

codecovBot commented Feb 23, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

Uh oh!

namrataghadi-galileo left a comment

Choose a reason for hiding this comment

Uh oh!

lan17 commented Feb 24, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

namrataghadi-galileo left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

lan17 commented Feb 23, 2026

Summary

Story

Validation

Uh oh!

codecovBot commented Feb 23, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

Uh oh!

namrataghadi-galileo left a comment

Choose a reason for hiding this comment

codecovBot commented Feb 23, 2026•
edited
Loading

lan17 commented Feb 24, 2026•
edited
Loading

codecovBot commented Feb 23, 2026•
edited
Loading

lan17 commented Feb 24, 2026•
edited
Loading

codecovBot commented Feb 23, 2026•
edited
Loading

lan17 commented Feb 24, 2026•
edited
Loading

codecovBot commented Feb 23, 2026•
edited
Loading

lan17 commented Feb 24, 2026•
edited
Loading