Skip to content

feat(setup): support Kimi Code client integrations - #1806

Closed
devloper961-maker wants to merge 1 commit into
morluto:mainfrom
devloper961-maker:feat/setup-kimi-code-client
Closed

devloper961-maker wants to merge 1 commit into
morluto:mainfrom
devloper961-maker:feat/setup-kimi-code-client

Conversation

@devloper961-maker

Copy link
Copy Markdown

Summary

Adds Kimi Code (\kimi_code) to setup client integrations, following the #1301 / #1409 pattern:

  • Registers the pinned stdio entry at \mcpServers.rea\ in user-level ~/.kimi-code/mcp.json, honoring a non-empty \KIMI_CODE_HOME\ override (used unexpanded)
  • Seeds \REA_MCP_INPUT_SCHEMA_PROFILE=compact\ in the written entry so every advertised schema stays under Moonshot's per-schema cap; explicit user values keep precedence and refreshes retain it
  • Detection via the ~/.kimi-code\ marker; shared ~/.agents/skills\ skill destination (no client-specific skill family needed)
  • Docs table row plus a Kimi Code note in \docs/installation.md; client ID added to the skill's connection-and-recovery reference

Test Plan


  • pm run verify:test-discovery\ — 827 files, new test discovered
  • New \ ests/boundary/filesystem/kimiClientConfiguration.test.ts\ — 7/8 pass; the 1 failure (package-runner alignment status) fails identically for Qwen on clean main in this Windows environment (pre-existing, environment-specific)
  • \ sc --noEmit\ clean; oxlint shows only the pre-existing max-lines warning; module boundaries check passes
  • \documentationFactIssues\ returns no issues against the built catalog (includes \kimi_code)
  • CI (Linux) to confirm the alignment/status assertions and full suite

Notes

  • No change to canonical tool contracts; the compact profile is presentation-only
  • User scope only, no project-level writes

Closes #1791

Register the user-level ~/.kimi-code/mcp.json entry (honoring a
non-empty KIMI_CODE_HOME override), seed
REA_MCP_INPUT_SCHEMA_PROFILE=compact so advertised schemas stay under
Moonshot's per-schema cap, and cover detection, registration, and
uninstall with boundary tests.

Refs morluto#1791
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 11, 2026 •

Copy link
Copy Markdown

Codex Review Summary

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

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-11T11:45:33.928925Z 3725960 PR opened
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 review" or "@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 chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 37259604fc

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

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

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

client.name === "kimi_code"
? Object.fromEntries(
Object.entries({
REA_MCP_INPUT_SCHEMA_PROFILE: "compact",

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Include Kimi's compact profile in the setup plan

When rea setup --client kimi_code --dry-run plans this write, SetupPlan.clientConfigurationDetail receives only providerEnvironment, so it reports “No additional provider environment variables” even though applying the plan adds this compact-profile variable. That makes the approved plan differ from the actual configuration mutation; pass the client-specific seed into the plan/detail so it is disclosed before approval.

AGENTS.md reference: AGENTS.md:L9-L9

Useful? React with 👍 / 👎.

client.name === "kimi_code"
? Object.fromEntries(
Object.entries({
REA_MCP_INPUT_SCHEMA_PROFILE: "compact",

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Validate Kimi's required profile in doctor

When a pre-existing Kimi registration has an owned REA command but neither its env nor the client process supplies REA_MCP_INPUT_SCHEMA_PROFILE=compact, readClientRegistrationStatuses still reports it as aligned because registrationAligned never examines this newly required setting. In that case scoped rea doctor --client kimi_code can report a healthy registration while Kimi sends the full schemas and Moonshot rejects every request; make the Kimi status stale when the profile seed is absent so doctor directs the user through setup.

Useful? React with 👍 / 👎.

@morluto

morluto commented Oct 11, 2026

Copy link
Copy Markdown
Owner

Thanks for putting this together. This overlaps with #1804, which was opened shortly before this PR and also implements Kimi Code setup with the compact schema profile while addressing #1791. To avoid maintaining two competing implementations, I’m closing this PR in favor of #1804. The connection-and-recovery skill reference update here is useful and does not appear in #1804’s changed files; please consider carrying that small documentation addition over there. Thanks again for the contribution.

@morluto morluto closed this Oct 11, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Feature] Support Kimi Code in rea setup client integrations

2 participants