feat(desktop): show the model's reported context window beside the declared setting

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J

[Joob1n]

Summary

Follows #4653, which is merged; this branch is a single commit on main.

After #4653 the context-window setting is the Maka window: the compaction target the user declares, and the only proactive threshold. A provider's /models report no longer takes effect on its own. This PR makes that visible where the user sets it: when nothing is declared and the connection's model catalog reports a window, the setting shows "This model declares a N-token window" with a one-click Use it that fills the declaration. The reported value never applies by itself. The help text now says what the setting does ("Maka compacts once the previous request's real usage exceeds it") and what leaving it empty means ("never compact proactively; the provider decides").

Refs #4559

Verification

npm --workspace @maka/desktop run typecheck, npm run check:renderer-architecture -- --base origin/main (no debt increase in provider-connection-detail.tsx), npm run astryx:surface-inventory, npm run lint, npm run format:check — clean. Not run locally: Desktop e2e — relying on CI.

Self-review

• The hint reads connection.models[].contextWindow (the Host's catalog entry) rather than models.dev directly, so it stays inside the dependency debt the ratchet allows for this legacy file.
• Applying the hint goes through the same setDraftContextWindow path as typing the value, so the declaration is persisted and validated exactly as a manual entry would be.

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Claude Code — implementation; reviewed and verified by the author.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

https://claude.ai/code/session_014ajaRxC4jydavY9nYUFj5J