architect-audit
Run the blueprint-validator against an existing blueprint and report PASS/FAIL with specifics. Blueprints written before v2 will FAIL — they have no acceptance criteria. That is expected, not a bug. / Audita un blueprint existente con el validador; los blueprints v1 fallan porque no tienen criterios de aceptación.
mkdir -p ~/.claude/commands && curl -fsSL https://raw.githubusercontent.com/Hainrixz/the-architect/HEAD/commands/architect-audit.md -o ~/.claude/commands/architect-audit.mdarchitect-audit.md
# `/architect-audit` — validate an existing blueprint
You are being invoked as `/architect-audit`. This command is **read-only**. It grades a blueprint;
it does not rewrite one.
**Say this before you dispatch anything, whenever the target has no `tasks.json` beside it:** a
blueprint written before v2 **will FAIL**, with a long finding list. It has no acceptance criteria at
all — that is the single thing the validator measures most of — plus no `tasks.json`, no epics, and
no verify commands. The failure is a diagnosis of a format gap, not evidence that the tool is broken
or that the design is bad. Setting that expectation up front is part of the job, because a
twenty-finding report delivered without it reads as an accusation.
The rubric lives in the validator, not here:
`${CLAUDE_PLUGIN_ROOT}/agents/blueprint-validator.md`. Read it before reporting, so your summary
uses the validator's own check names instead of invented ones.
## Resolve the target
`$ARGUMENTS` is a path. Handle these cases:
| Argument | Do |
|---|---|
| A `.md` file | Audit that file. |
| A directory | Glob it for the canonical bundle tree — `blueprint.md`, `tasks.json`, `epics/*.md`, and `workspace/` (`workspace/CLAUDE.md`, `workspace/AGENTS.md`, `workspace/.claude/`). Audit the bundle as a whole — a bundle missing `tasks.json` is itself a finding, and so is a `workspace/.claude/commands/` directory, which is never emitted. |
| Empty | Glob `./blueprints/*/` in the cwd, then `./blueprints/*-blueprint.md` for single-file mode. One match → use it. Several → list them and ask which. None → say so and point at `/architect`. |
| A path that does not exist | Say so. Do not guess a neighbor. |
## Run
Dispatch `blueprint-validator` with the `Task` tool, handing it the resolved path. Do not grade the
blueprint yourself — a second opinion from the same context is not a second opinion.
## Report
Print, in this order:
1. **Verdict** — `PASS` or `FAIL`, one line, no hedging.
2. **Findings table** — `Severity | Check | Where | What is wrong | Fix`. Blockers first. Point at
section headings or task ids, never "somewhere in the build order".
3. **Coverage** — how many build steps have an observable "Done when", how many have a verify
command, how many tasks in `tasks.json` have acceptance criteria. Report as `n/total`, not as a
vibe.
4. **Next move** — one line:
- Stale version pins only → `/architect-refresh <path>`.
- Missing acceptance criteria, no `tasks.json`, no epics → the structure is not there to patch;
re-run `/architect` and reuse the answers.
- Existing codebase → `/architect-brownfield`.
## Expect v1 blueprints to fail
Anything generated before v2 **will FAIL**. Not might — will. It has no acceptance criteria on any
step, no `tasks.json`, no epic files, and no verify commands, and those are most of what the
validator checks. A twenty-finding report on a v1 file is the tool working correctly.
So when the audited file has no `tasks.json` beside it, open the report with one line to that effect
before the verdict, and close the coverage block by naming the format gap rather than listing the
same absence twenty times. A v1 blueprint is still a usable design document. It is just not an
autonomously buildable bundle, which is the only thing this command measures.
## Rules
1. **Do not edit the blueprint.** If the user asks for fixes after seeing the report, that is a new
instruction — then use `/architect-refresh` for pin drift or `/architect` for structural gaps.
2. **No partial credit language.** "Mostly passes" is a FAIL with extra words.
3. **Every finding names a file and a location.** A finding you cannot locate is a hunch, and hunches
do not go in the table.
4. Detect and use the user's language.Adversarially audits a finished blueprint bundle and returns PASS or FAIL with line-referenced findings. Use before handing any blueprint to the user or to a build agent, and again after fixes. Read-only, Grep-driven, no shell. Fails on verify commands that reference files no build step creates, unobservable or machine-undecidable acceptance criteria, a migration with no Section 9.1 parity and cutover plan, missing sections, an empty Non-Goals scope fence, steps with no checkpoint tag, oversized steps, undocumented env vars, verify commands missing from the settings.json allowlist, dangling references, bad skill references, surviving placeholders, invented filenames for tool-generated artifacts, workspace files that are malformed or unignorable under the blueprint's own linter config (formatter *execution* is handed to the main thread's smoke test, not guessed at here), pins that imply verification that never happened, pins that no step ever installs, a step that retroactively breaks an earlier step's verify gate, an emitted runner config that cannot resolve a package the blueprint mandates, a standalone tool reading env vars nothing loads, an asserted count that disagrees with the blueprint's own content, checkpoint tags with no repository initialisation, an ignore file excluding a file the blueprint calls committed, two emitted artifacts that state the same path, entry point, name or port differently, an entry point that is built but never invoked, an emitted config that does not exclude the bundle's own path, a guard that exits non-zero on the path it guards against, a step Verify that asserts repository state only that same step's Checkpoint could produce, a byte-exact golden file or expected-output example that contradicts the blueprint's own data model or quotes a message only the pinned runtime could have produced, a gate whose pass condition is any non-zero exit so a usage error satisfies it vacuously, an ignore file or governing config delivered after the command it governs, and a tasks.json that does not match its epics. Triages pattern hits before filing them — an approval gate or a notarization command whose criterion resolves on this machine is correct work, not a finding.
Composes the finished blueprint from the interview findings, the chosen shape, the runtime track, and the selected capabilities — and in bundle mode writes the whole bundle: blueprint.md, tasks.json, epics/, and workspace/ (CLAUDE.md, AGENTS.md, .claude/, and the §19.6 verify-critical config files the gates need to run). Use after the architecture has been confirmed with the user, so the long generation runs in isolated context instead of flooding the interview thread. Its prompt must state the output mode. Returns the written paths, section coverage, an assumptions log, and any gap it refused to invent an answer for.
Resolves package versions and package identifiers against authoritative sources — the published artifact first, then the registry — before either is written into a blueprint. Use PROACTIVELY whenever a version is about to be pinned, a runtime track is refreshed, an export or option name is in doubt, or the user asks "what version should we use". Its report is authoritative over any cached runtime-track file. Returns package → current stable version → source URL → date checked, and flags prereleases, versions the rest of the stack cannot accept yet, unmaintained packages, and anything it could not verify.
Design a change against an existing codebase — maps the repo's stack, conventions, structure and tests, then emits a blueprint for a feature, refactor, or migration instead of a from-scratch build. / Diseña un cambio sobre un repo existente — mapea stack y convenciones y emite un blueprint de cambio, no de proyecto nuevo.
Resume a build — reads tasks.json from a bundle, finds the first pending task whose dependencies are all done, and prints it with its epic, acceptance criteria, and every verify command. This is what lets a long build survive across sessions. / Reanuda una construcción desde tasks.json y muestra la siguiente tarea desbloqueada.
Fast-track blueprint — three questions, smart defaults for everything else, still validator-gated. For when you just want it built. / Blueprint express — tres preguntas, defaults inteligentes, con el mismo validador.
Re-verify every pinned version in an existing blueprint against the live registries and report what moved, what breaks, and what to change — un-rots a months-old blueprint without a redesign. / Reverifica cada versión fijada en un blueprint contra los registros y reporta qué cambió y qué romper.
>-