Skip to content

docs: require exhaustive MCP-server use in project memory#120

Merged
div0rce merged 1 commit into
mainfrom
docs/mcp-exhaustive-usage
Jun 15, 2026
Merged

docs: require exhaustive MCP-server use in project memory#120
div0rce merged 1 commit into
mainfrom
docs/mcp-exhaustive-usage

Conversation

@div0rce

@div0rce div0rce commented Jun 15, 2026

Copy link
Copy Markdown
Owner

Summary

Adds a standing directive to the Local MCP/tooling memory section of CLAUDE.md and mirrors it identically into AGENTS.md: the configured MCP servers are the primary engineering surface for QSL and must be used exhaustively for every substantive task, rather than falling back to ad-hoc shell.

Task → server mapping made explicit: codescene review + analyze_change_set gate before PRs; git/github for diffs/blame/history and PR/issue/branch/commit/milestone review; qsl-results (SQLite) for benchmark history; codex for second-opinion review before finishing a milestone; context7 for current library docs; sequential-thinking, filesystem, docker. Practical guard kept: prefer the dedicated MCP, skip a server only when it genuinely cannot do the step (e.g. destructive git push stays manual per the golden rules); never invent an unconfigured server.

Also documents the four newly configured servers: codex, git, github, qsl-results.

Why a separate branch

This is a project-memory/policy change, unrelated to the in-flight M47 storage PR (#119), so it gets its own branch/PR to keep #119 scoped — matching the repo's precedent for memory-sync PRs.

Definition of Done

  • Directive added to CLAUDE.md and mirrored verbatim to AGENTS.md (the two sections differ only in the Claude Code client vs Codex client header line)
  • No code changed; docs-only
  • No benchmark or capability claims beyond what is configured
  • Human squash-merges

Notes

github MCP needs a one-time OAuth via /mcp; gh CLI remains the working fallback. Servers were installed at user scope (machine-local), so this PR only records the policy and roster in shared project memory.

Summary by CodeRabbit

  • Documentation
    • Enhanced guidance documentation with additional best practices for using configured tools and services.
    • Updated tool configuration documentation to reflect currently available integrations and capabilities.

Add a standing directive to the Local MCP/tooling section of CLAUDE.md
and mirror it identically to AGENTS.md: the MCP servers are the primary
engineering surface for QSL and must be used exhaustively for every
substantive task (codescene gate before PRs, git/github for review,
qsl-results for benchmark history, codex for second-opinion review,
context7 for any library API, sequential-thinking, filesystem, docker),
preferring the dedicated MCP over ad-hoc shell. Also document the four
newly configured servers (codex, git, github, qsl-results). Keeps the
two memory files in agreement per the repo's sync rule.

@codescene-delta-analysis codescene-delta-analysis 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.

No application code in the PR — skipped Code Health checks.

See analysis details in CodeScene

Quality Gate Profile: Pay Down Tech Debt
Install CodeScene MCP: safeguard and uplift AI-generated code. Catch issues early with our IDE extension and CLI tool.

@coderabbitai

coderabbitai Bot commented Jun 15, 2026

Copy link
Copy Markdown

Review Change Stack

Caution

Review failed

Pull request was closed or merged during review

📝 Walkthrough

Walkthrough

Both AGENTS.md and CLAUDE.md receive identical structural additions: a new "Local MCP/tooling memory" guidance block listing recommended MCP servers by task and a rule to prefer them over ad-hoc shell, plus four new entries (codex, git, github, qsl-results) appended to each file's configured-servers list.

Changes

MCP Tooling Documentation

Layer / File(s) Summary
MCP tooling memory block and server list expansions
AGENTS.md, CLAUDE.md
Both files gain a "Local MCP/tooling memory" section with per-server task guidance and a preference rule against ad-hoc shell work. Both configured-servers lists are extended with codex, git, github, and qsl-results entries.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Poem

🐇 Hop to the MCP, don't shell around!
Git, GitHub, codex — all servers found.
The docs say: reach first for the dedicated tool,
And qsl-results keeps benchmarks cool.
A rabbit reads docs and nods with a grin — ✅

🚥 Pre-merge checks | ✅ 4
✅ Passed checks (4 passed)
Check name Status Explanation
Title check ✅ Passed The title accurately summarizes the main change: requiring exhaustive MCP-server use in project memory documentation.
Description check ✅ Passed The description is substantially complete with clear summary, task-to-server mappings, rationale, and notes, though it omits several required template sections (Milestone, Tests, some DoD items).
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch docs/mcp-exhaustive-usage

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands and usage tips.

@div0rce div0rce merged commit 6ece05a into main Jun 15, 2026
7 of 8 checks passed

@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: 93fcfff39a

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".

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

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

Comment thread AGENTS.md
Comment thread AGENTS.md
div0rce added a commit that referenced this pull request Jun 15, 2026
Address PR #120 review:
- HANDOFF.md said 'use GitHub CLI over GitHub MCP', contradicting the new
  MCP-first directive (and resume reads HANDOFF after CLAUDE/AGENTS).
  Flip HANDOFF to one canonical workflow: prefer the GitHub MCP for
  PR/issue/branch/commit/milestone review, gh CLI as fallback when the
  MCP is unavailable or for scripted steps; likewise prefer context7 for
  current docs over memory.
- The directive named 'sequential-thinking' (hyphen) while the roster and
  HANDOFF use 'sequential_thinking' (underscore); standardize on the
  configured underscore name in CLAUDE.md and AGENTS.md so no line points
  at an unconfigured server name.
div0rce added a commit that referenced this pull request Jun 15, 2026
)

Address PR #120 review:
- HANDOFF.md said 'use GitHub CLI over GitHub MCP', contradicting the new
  MCP-first directive (and resume reads HANDOFF after CLAUDE/AGENTS).
  Flip HANDOFF to one canonical workflow: prefer the GitHub MCP for
  PR/issue/branch/commit/milestone review, gh CLI as fallback when the
  MCP is unavailable or for scripted steps; likewise prefer context7 for
  current docs over memory.
- The directive named 'sequential-thinking' (hyphen) while the roster and
  HANDOFF use 'sequential_thinking' (underscore); standardize on the
  configured underscore name in CLAUDE.md and AGENTS.md so no line points
  at an unconfigured server name.
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.

1 participant