菜单
Internal security review pass of the agentic-workflow review pack — composed in-turn by review-change and product-audit; not a menu entry. Checks secrets, input validation, injection, authn/authz, PII exposure, and dependency risk on the changed surface. Findings only; never edits code.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Periodic, product-wide health check — the CTO's "where do we actually stand?" Sweeps the WHOLE codebase (not a diff, not a PR) across every applicable axis — correctness, architecture, security/cybersecurity, performance, tests, UX/UI, accessibility, SEO, brand, tech debt — PLUS process & docs (incomplete phases, aging issues, solvable known-issues, doc/workflow completeness) and roadmap coherence. Mines accumulated suggestions from feature docs. Output: a severity-ranked report and concrete PROPOSALS — issues to open, roadmap features to add or remove. Every run is persisted to docs/audits/<n>-<YYYY-MM-DD>.md with an incremental audit id and findings numbered F1, F2, … so any finding is addressable later as "triage-issue <audit-id> F<k>". NEVER auto-fixes; the user decides what to act on. On Claude Code and want hand-tuned per-skill model/effort tiers? Install the `#claude` branch instead (`npx skills add gtrabanco/agentic-workflow#claude`) — see the README. This branch is model-agnostic: the skill inherits
Classify an issue and take a defensible decision: fix-now, postpone (deferred/trigger-based), wontfix, or promote-to-feature. Reads the issue's own "when to fix"/trigger and severity, verifies the trigger against the CURRENT codebase (counts consumers, checks thresholds, measures), then routes or reports with a dated, auditable comment. Also triages product-audit findings ("triage-issue <audit-id> F<k>"): reads docs/audits/<audit-id>-*.md, verifies the finding, and opens the GitHub issue only if the verdict warrants it. On Claude Code and want hand-tuned per-skill model/effort tiers? Install the `#claude` branch instead (`npx skills add gtrabanco/agentic-workflow#claude`) — see the README. This branch is model-agnostic: the skill inherits whatever model and effort your agent session is already using. Triggers: "triage issue N", "should we fix #N now", "classify this issue", "is #N's trigger met", "what do we do with #N".
Turn an idea or a feature request into an exhaustive, checkable product definition — the stage before engineering planning. The core mechanism is **capability closure**: a checklist that forces every entity, capability, and role a feature introduces to be walked to its full surface (CRUD + state transitions + UI entry point + API + test, or an explicit design-time `n/a`), so non-frontier executor models stop silently omitting the implicit work ("auth with dashboard management and ACLs" must not collapse to a users table + a list view). Closure has three fixed checklists: **entity closure** (each entity to full surface), **integration closure** (the feature reconciled against every subsystem in the project's capability inventory, `docs/CAPABILITIES.md` — auth, ACL, navigation, notifications, …), and the **role matrix** (every inventory role explicitly allowed/denied per capability) — plus an **expectation sweep** that forces what a competent human would implicitly assume ("a blog has drafts") into in-scope/out
Implement one phase of a feature (default), or a small feature / a fix (--fix) by its SPEC's ## Phases ledger — one phase per invocation, the final phase is always Hardening & PR (the close-out); legacy SPECs without ## Phases run end-to-end in a single pass. Enforces branch safety, issue policy, the project's verification gate, and per-phase doc discipline. On Claude Code and want hand-tuned per-skill model/effort tiers? Install the `#claude` branch instead (`npx skills add gtrabanco/agentic-workflow#claude`) — see the README. This branch is model-agnostic: the skill inherits whatever model and effort your agent session is already using. Triggers: "execute phase P1 of NN", "implement the NN feature", "build NN from its spec", "execute-phase NN P2", "execute-phase --fix".
Bootstrap a project's way of working: fetch the agentic-workflow documentation scaffold (template/) and adapt it to THIS project by interview — fill the CLAUDE.md documentation map, gate commands and architecture, prune doc folders that don't apply, keep the SPEC/feature/fix and GitHub templates — then offer to install the skills. The adaptive counterpart to a raw `npx degit` copy. On a repo that already has the scaffold, detects it and switches to **upgrade mode**: diffs the project's substrate against the current template, reads `docs/workflow/MIGRATION.md`, and proposes only the blocks the project is missing — never clobbering a tailored one. On Claude Code and want hand-tuned per-skill model/effort tiers? Install the `#claude` branch instead (`npx skills add gtrabanco/agentic-workflow#claude`) — see the README. This branch is model-agnostic: the skill inherits whatever model and effort your agent session is already using. Triggers: "set up the agentic workflow here", "init-workspace", "scaffold this proje
Internal step of plan-feature: turn a feature-request issue into a scoped, sized, roadmap-mapped SPEC **product half** (capability closure satisfied) with Closes #N traceability.
| name | review-security |
| user-invocable | false |
| version | 1.0.1 |
| author | Gabriel Trabanco <gtrabanco@users.noreply.github.com> |
| license | MIT |
| description | Internal security review pass of the agentic-workflow review pack — composed in-turn by review-change and product-audit; not a menu entry. Checks secrets, input validation, injection, authn/authz, PII exposure, and dependency risk on the changed surface. Findings only; never edits code. |
Composed by review-change / product-audit within their conversation — on any
agent, follow this file inline as the routed step. Findings only; never edits,
never refactors.
The diff or path/glob the caller passes; default the current change vs the default branch. State the scope at the top of the returned table.
✓ No secrets/credentials/tokens in code, config, tests, or fixtures (grep the diff for key-like strings) ✓ Every external input on the changed paths is validated/sanitized before use ✓ No injection vectors (SQL/command/path/template) — parameterized/escaped, never concatenated ✓ AuthN/AuthZ enforced on every new/changed endpoint or entry point (cite where) ✓ No PII or secrets written to logs/error messages on the changed paths ✓ Webhooks/callbacks verify signatures before processing ✓ Rate limiting / abuse controls considered where a new public surface appears (n/a if none) ✓ New/updated dependencies pinned and free of known-critical advisories (state how you checked) ✓ Error responses don't leak stack traces or internal paths ✓ Unsafe deserialization / dynamic evaluation of untrusted data absent
REVIEW SECURITY — scope: <scope>
| # | Finding | Sev | Evidence | Suggested fix |
|---|---------|-----|----------|---------------|
| 1 | <what> | critical|major|minor | <file:line> | <smallest action> |
Checklist: <n> evaluated, <n> pass, <n> findings, <n> n/a (<which + why>)
Summary: <1-2 sentences>
Decision: PASS | FAIL
FAIL if any critical or major finding is open; PASS otherwise. Minor findings never block — they route to the caller's triage step.