q-validate

Q-Validate — multi-round plan review

Completion gate

For every council start, q_await is the blocking completion notification. After completing the host's independent work, call it in the same turn as the start. Do not end the turn, tell the user you will check later, or leave a live job_id pending. If the council is still running, remain in the blocking call until it returns a result or error.

Arguments

$ARGUMENTS is the path to the plan file, optionally with --verbose to lift the terse output caps and/or --gemini-model <id> to run the gemini seat on a specific agy model. Strip --verbose and --gemini-model <id> from the path before passing plan_path. If the remaining path is empty, ask the user which plan file to review.

Host routing

Claude Code uses the mcp__plugin_code-quorum_quorum__* tools and passes host: "claude". Codex uses the mcp__quorum_codex__* tools and passes host: "codex". The workflows are otherwise identical.

If either required MCP tool is absent from the current task's tool registry, stop and report that the plugin MCP server did not load. Do not substitute the synchronous CLI, direct Python imports, or another runtime path: only the registered start/await pair enforces the structural anti-bias gate.

Every MCP council start requires an explicit absolute cwd. Never omit it: the MCP server's own working directory is its installation or runtime location, not the project selected by the user.

Step 1 — Start the agents (non-blocking)

Call mcp__plugin_code-quorum_quorum__q_validate_start (Claude Code) or mcp__quorum_codex__q_validate_start (Codex) with:

• plan_path: a bare file path only (absolute, or relative to cwd) that resolves inside cwd; paths outside the working directory are rejected before reading. q_validate_start open()s this path directly, so it CANNOT carry an appended context/boundaries block — passing <path> + prose errors with File name too long. If $ARGUMENTS includes reviewer context beyond the path, do NOT forward it here; instead put that context, boundaries, or specific questions inside the plan file (e.g. an ## Open questions for review section) — the agents read the file.
• cwd: the absolute path of the project working directory
• host: the current host selected above
• extended: false by default (2-round review). Set true for a 4-round deliberation with a stance rotation at round 3 — use only when the plan is contentious or high-stakes.
• mode: "revise" (default) — agents soften or strengthen positions in light of peers. Use "critique" when you want them to attack each other's points.
• agents: optional; omit it for every external seat in the current host profile (a seat the user has marked disabled in quorum setup-models is skipped from that default roster). Passing agents explicitly overrides the skip and runs a disabled seat anyway — a deliberate ask always runs.
• verbose: false by default. The council writes terse output — no prose padding, path:line citations instead of pasted code, and later rounds omit restating what is unchanged. Set true only if the user gave --verbose; it produces a much larger matrix.
• gemini_model: omit by default. Pass an agy model id (exactly as printed by agy models) when the user gave --gemini-model <id> or asked for a specific model in the gemini seat — e.g. claude-opus-5-5-high to have that seat answer with Claude on the same AI Pro plan.