Address remaining SDK-288 tracing UX issues by Qard · Pull Request #39 · braintrustdata/braintrust-coding-agent-plugins

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Stephen Belanger (Qard) · 2026-09-01T14:34:28Z

Summary

add CLI-friendly, fully specified parent identifiers for trace imports while retaining SDK span exports
print structured import summaries in human and JSON output
reject Codex's incompatible global hook-trust bypass with an actionable explanation
warn about obsolete Claude tracing env keys without changing shared Claude settings or removing BRAINTRUST_API_KEY

Addresses SDK-279, SDK-280, SDK-282, and SDK-284.

Validation

make test
cargo test --manifest-path bt-daemon/Cargo.toml --all-features
cargo clippy --manifest-path bt-daemon/Cargo.toml --all-features --all-targets -- -D warnings

chatgpt-codex-connector · 2026-09-01T16:06:57Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed2026-09-01T16:11:30.740540Z	`0317e8e`	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:0317e8ef77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

## Summary Implements the `bt`-owned SDK-288 work: - SDK-272: make `trace enable` / `trace setup`, `trace run`, and `trace import` complete from credentials already available to the command. `BRAINTRUST_API_KEY` remains invocation-local for run/import; persistent setup materializes it into a normal saved profile for future hooks; interactive commands collect a missing key, org, or project instead of requiring a separate login step. - SDK-274: require explicit consent before ordinary `bt login` saves `BRAINTRUST_API_KEY`; persistent tracing setup itself is the explicit request to create the durable login it needs. - SDK-270 / SDK-275: keep bare `bt status` concise; expand saved profile and credential-storage details only under `--all`, and show precedence only for an active environment override. - SDK-276: clarify that logout is local-only and add a one-confirmation `logout --all` flow with per-profile JSON results. Profile deletion clears config references before removing metadata, and rename stages both names while references move so config never points to a missing profile. - SDK-273: provide side-effect-free, credential-redacted auth provenance to `bt trace doctor`. - SDK-277: add `bt profiles doctor` and `bt profiles repair`. Automatic diagnosis and cleanup operate on the enumerable plaintext fallback store; explicit deleted profile names can be repaired across the OS credential store and plaintext fallback without unsafe keychain enumeration. - SDK-278: format no-browser/remote OAuth as labeled authorization, callback, credential-storage, and logged-in-profile sections while keeping JSON and quiet output clean. - SDK-281: advance the embedded daemon to the integrated plugin-monorepo `main` commit containing PR #39 and stable trace profile IDs. ## Dependency braintrustdata/braintrust-coding-agent-plugins#39 has been merged. This PR pins plugin-monorepo `main` at exact commit `202b3e305ea7ff74bc6ea117f46a32cc18b61b25`, which contains both the PR #39 import/setup fixes and the subsequent stable-profile-ID migration from #44. ## Validation - `cargo fmt --check` - `cargo clippy --all-targets --locked -- -D warnings -A clippy::collapsible_match` - `cargo test --all-targets --locked -- --skip eval::tests::eval_args_from_env_populates_supported_fields --skip eval::tests::eval_args_parse_first_sampling_flag --skip eval::tests::eval_args_parse_max_concurrency_flag` - 793 unit tests passed; 3 known eval argument-parser tests skipped - 62 CLI tests passed - dataset, eval dev-server, eval fixture, functions, and setup integration suites passed - focused regression proving environment-auth setup persists and routes by the new stable profile ID - `cargo check --locked` - `git diff --check` The three excluded eval tests fail without these changes because their parser inputs combine `--dev` with `--max-concurrency`, which clap declares mutually conflicting. Current Rust 1.95 clippy also reports unrelated `collapsible_match` findings in existing dataset, SQL, and trace code; the repository's existing lint invocation allows that lint.

Address remaining tracing import and setup issues
0317e8e

This was referenced Sep 1, 2026
Fix SDK-288 tracing auth and diagnostics braintrustdata/bt#334
Merged
Consolidate coding-agent tracing behavior #26
Closed

Stephen Belanger (Qard) marked this pull request as ready for review September 1, 2026 16:06

Stephen Belanger (Qard) requested a review from Abhijeet Prasad (AbhiPrasad)September 1, 2026 16:06

Stephen Belanger (Qard) requested a review from Andrew Kent (realark)September 1, 2026 16:07

chatgpt-codex-connectorBot reviewed Sep 1, 2026
View reviewed changes

Comment threadbt-daemon/src/lib.rs
Comment threadbt-daemon/src/paths.rs Outdated

Address PR review feedback
0975ff1

Stephen Belanger (Qard) self-assigned this Sep 1, 2026

Abhijeet Prasad (AbhiPrasad) approved these changes Sep 1, 2026
View reviewed changes

Stephen Belanger (Qard) merged commit 3409ef7 into mainSep 1, 2026
20 checks passed

Stephen Belanger (Qard) deleted the t3code/sdk-288-followups branch September 1, 2026 16:45