[kixelated]

Problem

moq-dev/quest#457 added quest context, which prints shared engineering rules plus a Prompting section, and moq-dev/quest#460 made it the command agents run before any work. The studio still carried its own copies of those rules in AGENTS.md and PROMPTING.md.

Approach

• Bump the quest flake input to moq-dev/quest@4be3508. The input pins a rev in flake.nix, so the bump edits that URL as well as flake.lock.
• Trim root AGENTS.md to the studio-specific rules (Upstream, Rust, Development, perf metrics). Its reference line is quest's default, Quests: before any work, run `quest context` and follow it., and the next line gives the nix develop --command quest ... fallback for running quest outside the dev shell.
• Delete PROMPTING.md; .claude/rules/prompting.md and REVIEW.md point at Prompting in quest context.
• REVIEW.md lists quest context as a rules source and tells the reviewer to run it, since quest review passes the reviewer only the base's AGENTS.md and REVIEW.md. A reviewer that cannot run it says so in its review instead of reviewing silently without the rules.

Every dropped AGENTS.md rule is covered by quest context or quest guide. A few generic rules have no counterpart there and were dropped as not studio-specific: "no unrelated refactors or formatting churn", "match existing conventions", and "don't pass a flaky test by loosening its assertion".

Impact

Docs and tooling only: agent instructions and the dev shell's quest binary. No public API or wire change.

Alternatives

• Keep a copy of the shared rules in REVIEW.md (or AGENTS.md) so a reviewer that cannot run quest context still gets them. Rejected: it restores the duplication this PR removes, and the gap is closed at its source by quest review passing the context (see Follow-ups).
• Wait for that upstream change before landing. Not needed: every studio review so far ran Codex, which runs quest context in its read-only sandbox (checked with quest review's Codex flags).

Follow-ups

• moq-dev/quest: quest review should pass assets/CONTEXT.md to the reviewer, as src/runner/prompt.ts already does for jobs. Under --permission-mode dontAsk with no settings, the Claude Code reviewer is denied quest context (checked with quest review's Claude flags), so a Claude review of a Codex-authored PR lacks the shared rules until then. Once it lands, bump quest here and drop the REVIEW.md line telling the reviewer to run it.

Validation

• nix develop --command quest context prints the context.
• nix develop --command just ci passes on 06a7bbe.
• No tracked file outside moq/ still references PROMPTING.md.

Review

• Reviewer: Codex (quest review --base origin/main), model gpt-6.1-sol, effort medium
• Reviewed head: 06a7bbe
• Round 1 (new session): no findings

🤖 Generated with Claude Code

(Written by Claude Opus 5.5)

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 30 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 71af2f93-25e3-49a7-a9a7-25a02ef17051

Commits

Reviewing files that changed from the base of the PR and between 720d021 and a863061.

Files ignored due to path filters: 1

• flake.lock is excluded by !**/*.lock

Files selected for processing: 5

• .claude/rules/prompting.md
• AGENTS.md
• PROMPTING.md
• REVIEW.md
• flake.nix

Walkthrough

Repository guidance now directs contributors to use quest context and its Prompting guidance. The root PROMPTING.md file is deleted, and REVIEW.md and the prompting rule are updated to refer to the new guidance source. AGENTS.md also removes general instructions and changes several development instructions. The quest flake input points to a different revision.

Merge Risk: Low · previous review 720d0

The documentation migration leaves agents without access to Prompting guidance when the development shell fails to load. Provide a documented fallback; the impact is bounded to contributor workflows.

Pre-merge checks: 4

Check name	Status	Explanation
Title check		The title names the main change: agent instructions now rely on quest context for shared rules. The changeset matches this. It edits root AGENTS.md, deletes PROMPTING.md, repoints .claude/rules/prom…
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.
Description check		The pull request description directly explains the changes to quest integration, agent instructions, PROMPTING.md, REVIEW.md, and validation. It matches the changeset.

Full details: Title check

Explanation

The title names the main change: agent instructions now rely on quest context for shared rules. The changeset matches this. It edits root AGENTS.md, deletes PROMPTING.md, repoints .claude/rules/prompting.md and REVIEW.md to quest context, and bumps the quest flake input. The title uses the conventional docs(agents) scope and stays concise.

Finishing Touches

Simplify code

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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

• X
• Mastodon
• Reddit
• LinkedIn

Recent review info

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 9bfeacd9-acde-45d1-a79f-5054ec05081d

Commits

Reviewing files that changed from the base of the PR and between a05d702 and 720d021.

Files ignored due to path filters: 1

• flake.lock is excluded by !**/*.lock

Files selected for processing: 5

• .claude/rules/prompting.md
• AGENTS.md
• PROMPTING.md
• REVIEW.md
• flake.nix

Files with no reviewable changes: 1

• PROMPTING.md

Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.

[kixelated]

Automated review by review (OpenAI)

Reviewed head: 657cf41

No actionable findings. I compared the removed shared rules and prompting guidance with the context and guide bundled by the pinned quest revision, and checked the flake and shell wiring. The intentionally dropped generic preferences are identified in the PR description.

Verification: GitHub source review only; I did not run Nix, the CLI, or the checks. The reported successful commands are author-reported.

(Written by review (OpenAI))

[pullfrog]

Important

The trim makes the reviewer's shared rules depend on quest review handing them over, which it does not do today.

Reviewed changes

• Trimmed AGENTS.md: root file now keeps only the studio-specific Upstream, Rust, Development, and perf rules, delegating shared rules to quest context and quest guide.
• Deleted PROMPTING.md: .claude/rules/prompting.md and REVIEW.md now point at Prompting in quest context.
• Repointed REVIEW.md: lists quest context as a rules source and drops the AGENTS.md-specific wording in "Not findings".
• Bumped the quest flake input: flake.nix and flake.lock move quest to 9277dc3, which also advances the transitive quest-pinned input.

ℹ️ Three guardrails are dropped instead of upstreamed

The body says "no unrelated refactors or formatting churn", "match existing conventions", and "don't pass a flaky test by loosening its assertion" have no quest context counterpart and were dropped. They are repo-agnostic, so the durable home is quest's CONTEXT.md, not nowhere; dropping them locally loses them for studio agents. The first also now sits against the retained quest context rule "Refactor aggressively for long-term maintainability", which can encourage exactly the out-of-scope refactors the deleted rule forbade.

Technical details

# Dropped guardrails belong in quest context

## Affected sites
- `AGENTS.md` (deleted lines): "Keep the PR to one problem: no unrelated refactors, formatting churn, or drive-by changes."
- `AGENTS.md` (deleted line): "Match the existing conventions, patterns, and naming when possible."
- `AGENTS.md` (deleted clause): "nor pass a flaky test by loosening its assertion or disabling production behavior."

## Required outcome
- Each of these rules still reaches studio agents through some always-loaded file, ideally upstream in `moq-dev/quest`'s `assets/CONTEXT.md`.

## Open questions for the human (optional)
- Is upstreaming these three to quest the intended follow-up, or should the studio keep a local copy until it lands?

  ^{｜ Fix all ➔ ｜ Fix 👍s ➔ ｜ View workflow run ｜ Using big-pickle (free) ｜ 𝕏}

[kixelated]

Automated review by review (OpenAI)

Reviewed head 720d021 after 657cf41.

The new Quest pin moves the general workspace, question, and review guidance into quest context, and the updated reference directs agents to it before work. I found no additional actionable issue in this delta. The existing Pullfrog thread on whether the review harness actually receives those rules remains open; the pin update does not itself change quest review's instruction injection.

Verification: static GitHub inspection of the Studio delta and the pinned Quest sources only; I did not run Nix, Quest, or a reviewer session.

[pullfrog]

ℹ️ No blockers. One documentation inconsistency in the update note.

Reviewed changes

This run re-reviewed the push that bumped the quest pin and narrowed the always-run command.

• Narrowed the quest line in AGENTS.md: it now says to run only quest context before any work, dropping the separate quest guide step.
• Bumped the quest flake input to 4be3508: at that rev quest context carries the workspace, questions, pull-request, review, and quest rules the guide used to hold, so the dropped quest guide step is still reached for quest work.

ℹ️ CONTRIBUTING.md was not repointed at the Reviews rules

The update note says this push "repoints CONTRIBUTING.md references to the moved Reviews rules", but CONTRIBUTING.md is untouched by the whole PR. It now nearly restates the Reviews section that quest context gains at the new pin, so the consolidation this PR is for is missing where the shared review workflow lives.

Technical details

# Repoint CONTRIBUTING.md's Reviews section, or drop the claim

## Affected sites
- `CONTRIBUTING.md:43-49` - the `# Reviews` section repeats the ready-for-review review, the `## Review` record contents, and "address posted findings, never request one" that `quest context` carries at `4be3508`.
- `720d021` - touches only `AGENTS.md`, `flake.nix`, and `flake.lock`; no hunk of this PR touches `CONTRIBUTING.md`.

## Required outcome
- The PR body stops claiming a `CONTRIBUTING.md` change that is not in the diff, or `CONTRIBUTING.md` keeps only what is studio-specific (`/quest-review`, the `## Review` record, `## Waiting` handling) and lets `quest context` carry the mechanics.

## Open questions for the human (optional)
- Was the `CONTRIBUTING.md` trim meant for this push, or is the update note stale?

  ^{｜ Fix it ➔ ｜ View workflow run ｜ Using big-pickle (free) ｜ 𝕏}

[coderabbitai]

• Fix comments: Fix CodeRabbit comments on this PR

[kixelated]

On pullfrog's review notes:

• CONTRIBUTING.md claim: the update note was wrong; CONTRIBUTING.md had no quest guide references to repoint. The rewritten body drops the claim. Its Reviews section stays as the studio's entry point (/quest-review, the ## Review record).
• Dropped guardrails: dropping them here was the author's choice, recorded in the body. If they should survive, their home is quest's assets/CONTEXT.md, not a studio copy, so they are a suggested upstream follow-up rather than restored here.

(Written by Claude Opus 5.5)

[kixelated]

Automated review by review (OpenAI)

Reviewed head: 06a7bbe, following 720d021.

The added REVIEW.md instruction addresses the earlier visibility gap for reviewers able to run quest context, and explicitly asks reviewers to disclose when they cannot. The PR description identifies the remaining Claude permission-mode limitation and upstream harness follow-up. I found no additional actionable issue in this change. The base merge brings in the failover feature without changing these instruction files.

Verification: static GitHub inspection only; I did not run Nix, Quest, or reviewer sessions, so the reported behavior of those sessions is author-reported.

(Written by review (OpenAI))

[pullfrog]

ℹ️ No blockers. One suggestion inline about where the reviewer reads CONTRIBUTING.md from.

Reviewed changes

Re-reviewed the push since the last Pullfrog review (720d021), which is 06a7bbe plus merges that only sync main.

• Added a nix develop fallback in AGENTS.md: the quest line now tells agents to run nix develop --command quest ... outside the dev shell, so they still reach quest context without direnv.
• Directed reviewers to fetch the rules themselves in REVIEW.md: since quest review injects only the base's AGENTS.md/REVIEW.md, reviewers must run quest context and read CONTRIBUTING.md, and say so when quest context cannot run.

  ^{｜ Fix all ➔ ｜ Fix 👍s ➔ ｜ View workflow run ｜ Using big-pickle (free) ｜ 𝕏}

[pullfrog]

✅ No new issues found.

Reviewed changes

Re-reviewed the push since the prior Pullfrog review (a404ed9), which is a863061.

• Sourced the reviewer's CONTRIBUTING.md from the base in REVIEW.md: the line now says to read "the base's CONTRIBUTING.md" and gives git show <base>:CONTRIBUTING.md, so a PR can no longer edit the rules that judge it. This closes the one inline finding from the prior review, and both open Pullfrog threads are now resolved.

  ^{｜ View workflow run ｜ Using big-pickle (free) ｜ 𝕏}

[kixelated]

Automated review by review (OpenAI)

Reviewed head: a863061, following 06a7bbe.

No new actionable findings. Reading CONTRIBUTING.md from the base prevents a PR from changing that review guidance for its own review. The earlier Quest execution limitation remains as described in the PR; this change does not alter the review harness.

Verification: static GitHub inspection only; I did not run Quest or a reviewer session.

(Written by review (OpenAI))