review-plan-security

Review security implications of the implementation plan via Codex (round-continuing under the 2+1 cap: fix concerns and re-run).

Procedure

RPS-1. Resolve <PLANS_DIR> via skills/_shared/resolve-plans-dir.md. Identify <PLANS_DIR>/<session-id>-detail.md. Initialize EXTENSIONS_USED=0. RPS-2. Invoke "$AGENTS_MAIN_ROOT/skills/review-plan-security/scripts/run-codex-review-loop.sh" (Bash), exporting AGENTS_MAIN_ROOT, SESSION_ID, PLANS_DIR, EXTENSIONS_USED. Pass CTX_SURVEY_CODE, CTX_SURVEY_HISTORY as env vars when available. CTX_CONCERNS_LOG is auto-generated by the wrapper from the ledger (concern-ledger render-concerns-log, format security-plan) and auto-exported when non-empty — do not set it manually (#2185). Exit-code handling (SSOT: skills/_shared/codex-review-loop.md; round-continuing under the 2+1 cap):

• exit 0 APPROVED → RPS-5 (no RISK items).
• exit 1 NEEDS_REVISION → save stdout to <PLANS_DIR>/<session-id>-security-plan-codex-round-<N>-raw.md (<N> from <CONTROL_DIR>/security-plan-last-round.txt, <CONTROL_DIR> = node "$AGENTS_MAIN_ROOT/bin/workflow-control-dir" --session <session-id>); present concerns via RPS-3; address them, then re-run /review-plan-security — the round counter survives, so the re-run is counted as the next round.
• exit 2 ESCALATE → run review-loop-summarize-concerns --budget-remaining 0; stop → RPS-3.
• exit 6 HIGH_UNRESOLVED → save stdout to <PLANS_DIR>/<session-id>-security-plan-codex-round-<N>-raw.md (<N> from <CONTROL_DIR>/security-plan-last-round.txt); run review-loop-summarize-concerns --budget-remaining 0; stop → RPS-3. Re-run only after accepting per exit-codes.md "Escalation by format".
• exit 3 → silently launch plan-security-reviewer subagent with model: = the model= line of node "$AGENTS_MAIN_ROOT/bin/resolve-role-model" --role reviewer; its APPROVED verdict → RPS-5; its NEEDS_REVISION verdict → RPS-3. Its prompt carries the canonical line of the rules/shell-commands.md Subagent Dispatch Prompts section.
• exit 4 → HALT with blocking error; surface wrapper stderr; do NOT launch fallback agent.
• exit 5 AUTO_EXTEND → same as exit 1 (concerns remain within the extension budget): present via RPS-3, address, re-run.
• exit 7 FINALIZE_FAILED → <CONTROL_DIR>/security-plan-unresolved-concerns.json could not be written; HALT, surface the ## Concern Ledger: FINALIZE-FAILED line, launch no fallback, emit no sentinel. After an ESCALATE, confirm the artifact with bash "$AGENTS_MAIN_ROOT/bin/concern-ledger" check-finalized --plans-dir <PLANS_DIR> --session-id <session-id> --format security-plan first.
• exit 9 → HALT; accept per exit-codes.md "Escalation by format", then re-run. RPS-3. Triage — per skills/_shared/priority-hierarchy.md, reject a concern only when it directly contradicts a decision already settled in the approved intent.md / outline.md (including their ## Accepted Tradeoffs).
• Raising a topic the plan does not address is never grounds to reject.
• Rejecting obliges naming the specific governing decision; a reject carrying no cited decision is a procedure violation.
• Record each reject durably so it does not re-surface next round: bash "$AGENTS_MAIN_ROOT/bin/concern-ledger" reject --plans-dir <PLANS_DIR> --session-id <session-id> --format security-plan --id C<N> --reason "<cited decision>" (#2185).
• A concern about the consequences or residual risk of a settled decision is no contradiction of it; only a demand to reverse the decision is.
• Forward each surviving concern to RPS-4.
• When every concern is rejected on a cited decision, treat the review as having no RISK items and go to RPS-5.
• exit 0 carries no concerns and skips RPS-3. RPS-4. Present the surviving concerns with per-axis severity and proposed mitigations before implementation proceeds. RPS-5. Summary — APPROVED (including the all-rejected case): report no RISK items, and on the all-rejected path also list every rejected concern beside the settled decision cited against it; NEEDS_REVISION: summarize the RPS-4 mitigations.