inspire-workspace
Workspace-level validation: the global pre-PR review (orchestrating module + cross-module checks) and vault-structure validation. Use for the required pre-merge gate and top-level structure checks. ADR lifecycle lives in /inspire-adr; the task tracker in /inspire-task.
- License 许可
- MIT
- 最近更新
- 2026-11-13
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- inspire-workspace
- description
- Workspace-level validation: the global pre-PR review (orchestrating module + cross-module checks) and vault-structure validation. Use for the required pre-merge gate and top-level structure checks. ADR lifecycle lives in /inspire-adr; the task tracker in /inspire-task.
02 资源文件 3
- 参考文档references/workspace-review.md6.5 KB
- 参考文档references/workspace-structure.md5.1 KB
- 脚本review.workflow.mjs6.0 KB
03 SKILL.md 正文
/inspire-workspace — Workspace-level Operations
Scope
This skill owns workspace-scoped validation — the cross-cutting checks that don't belong to a single module or feature:
- Global review — the pre-merge gate, orchestrating module-level and cross-module checks.
- Vault structure — the
01_spec/layout, typed extensions, and the tracker's on-disk invariants.
It does not own the artifacts it validates the coherence of: ADR lifecycle
is /inspire-adr and the task tracker is
/inspire-task. This skill reads ADRs and tickets to
judge coherence; it never authors them.
Several checks in review and structure read against the project's faces
(*.face.md) and bundles (*.bundle.md). What faces: and bundles: mean on an
artifact, and how scope resolves, are defined once in
_references/face-scope.md; read it before judging
anything face- or bundle-scoped. A project without screens has no face, and the
face-scoped checks have nothing to read.
Invocation
/inspire-workspace review— full vault review/inspire-workspace review {module1} {module2}— scoped to selected modules + cross-module checks/inspire-workspace structure— validate the01_spec/layout, typed extensions, task tracker, vault conventions
Subcommands in references/
Each subcommand's core procedure lives in a reference file. Before executing any subcommand, read every reference file its index row names — the table below is an index, not the flow.
| Subcommand | What its reference holds |
|---|---|
review |
the validators, Phases 3–6 + Signals — the cross-cutting checks; the Execution mode, Phases 1–2, the Output format skeleton and the ### Review rules stay in this file; Phase 2 additionally reads inspire-module/references/module-review.md |
structure |
The full top-level vault checks + their output skeleton |
Subcommand: review (global)
REQUIRED before any PR that modifies files in inspire_kb/. Orchestrates a
full consistency review across all affected modules and the vault structure.
Execution mode
The checks, severity model, and output are identical in both modes — only the scheduling differs.
- Sequential (default). Execute the phases top-to-bottom in this agent — Phases
1–2 below, then the validators, Phases 3–6 + Signals in
references/workspace-review.md. - Workflow (opt-in, when the user enables ultracode / multi-agent). After
Phase 1 (scope), run the bundled workflow at
review.workflow.mjsvia the Workflow tool — passargs: { modules: [<in-scope slugs>] }and preferscriptPathwith that file's absolute path. It fans the per-module reviews out in parallel, runs a deterministic completeness gate (any dropped module review becomes acritical: review-incompletefinding — never a silent pass), then a single synthesizer runs the cross-cutting phases over the full repo and emits the standard report.
Invariants both modes MUST preserve: read-only (flag, never edit, never invoke
a fix-skill); ADR-propagation alignment is judged by reading each ADR's
Status + Decision — design-workspace coherence (features + screen spec + horizontal
prototype + specs) is required at every maturity, and higher maturities add
external evidence checkable only by pointer (prototyped → a **Prototype:**
pointer to an external functional prototype; implemented → a codebase reference);
and the output uses the exact skeleton in Output format below.
Phase 1 — Identify scope
- If modules are specified, use those. Otherwise, find every
*.module.mdunderinspire_kb/01_spec/: each one is a module, and its id is the file name without.module.md. - For each module in scope, delegate to
/inspire-module review {module}.
Phase 2 — Module reviews
For each module in scope, the module review performs the checks in
inspire-module/references/module-review.md
— referenced, never restated.
Output format
# Global Review | {date}
## Scope
- Modules reviewed: {list}
- Validators: `review.sh inspire_kb` — {PASS | N errors, M warnings}
## Summary
{X} issues: {critical} critical, {important} important, {minor} minor
Drift items pending: {N}
## By Module
### {module}
- [{severity}] {description} — {file}:{line} | Fix: `/{skill}`
## Cross-Module
- [{severity}] {description}
## Vault Structure
## Prototype Component Adoption
- {Component}: {adopted}/{total} pages
- {Component} — {shell}: {adopted}/{total} pages (one line per shell, from 2+ faces)
## Catalog Coherence
- Patterns: {total} defined, {used}, {orphan}
- Components: {total} defined, {used}, {orphan}
- Design system variance: reported under Signals
## Signals
{`.inspire/bin/trust.sh report` output, verbatim}
- Design system: {N} per-face variant section(s), {L} lines
## OK
Review rules
- Be thorough. This is the pre-merge gate.
- Be specific. Every finding includes a file path + line number.
- Prioritize by impact. Critical = broken refs, missing files, a module review
that did not complete (
review-incomplete, see Execution mode), ADR consequences not reflected within their maturity's reach. Important = stale content, missing coverage, legacy structure. Minor = naming, formatting. - No false positives. If unsure, note as "verify".
- Actionable. Every finding suggests the skill to invoke.
- Delegate deep dives. For complex feature-level issues, suggest
/inspire-feature review {id}. - Pending drift is not failure. Prototype drift and pending component adoption
are
important, notcritical, unless they contradict an ADR within its maturity's reach. - An undeclared contradiction is drift, not an override. Where two layers
answer one question differently, the narrower artifact wins only where it
declares the departure in the open; otherwise the wider rule holds and the
contradiction is a finding routed to the narrower artifact's owning skill
(
_references/authority.md). Report the same departure declared in a third artifact as a decision sitting at the wrong rung, and name the skill that would move it —/inspire-adrfor the product,/inspire-lessonfor the runtime. This does not soften rule 7: nothing below the decision rung has a place to declare an ADR departure, so contradicting a current ADR within its maturity's reach is never an override and escalates exactly as rule 7 says. - Signals are measurements, not findings. This is the single statement of how
every signal is treated — the trust report's groups and the design-system
variance count alike. They are reported every run, including when nothing is
wrong; they carry no severity of their own, and never above
importantwhere one is shown; no fix routing beyond the owning skill; never file a ticket from one; never block the pre-PR gate on one. They re-appear for as long as they stay true (trust-stamps). - Consult the task tracker. Known items in
inspire_kb/99_tracker/tickets/are flagged(tracked: TASK-{id}). Use/inspire-task list. - Required follow-up skills. When flagging drift, name the mandatory fix skill:
- Prototype drift →
/inspire-prototype - screen spec drift →
/inspire-screens - Feature drift →
/inspire-module updateor/inspire-feature update - ADR misalignment →
/inspire-adr update - A face or bundle finding, or an unresolved
faces:/bundles:id →/inspire-face review - A process finding →
/inspire-process review
Subcommand: structure
The full procedure — the top-level vault checks and their output skeleton —
lives in references/workspace-structure.md.
Rules
Output language. Write review reports and findings in the project's declared
output_language(default English), per_references/output-language.md. Applies whatever language the conversation is in; machine-read tokens (frontmatter keys/values, link paths, filenames) stay verbatim.
Writing contract. Review reports and findings follow
_references/writing-style.md.
Lesson capture. At a natural pause, when the operator's feedback should change how this skill behaves, offer
/inspire-lesson note— never auto-write a lesson. Protocol and ticket-vs-lesson routing:_references/lesson-capture.md.
reviewandstructureare read-only. They suggest and flag; they never edit files or invoke a fix-skill.- Consult the task tracker. Known items live in
inspire_kb/99_tracker/tickets/(/inspire-task list); don't re-report them as new findings. - Git discipline is shared. Branch/commit/PR conventions, merge-conflict
auditing, and the git safety protocol live in
_references/git-conventions.md(sensible defaults; the project'sCLAUDE.mdoverrides). Follow it whenever the operator asks for a branch, commit, or PR — and never commit/push on your own initiative.
Related skills
/inspire-adr— ADR lifecycle (create / update / promote / supersede). Review judges whether ADR consequences propagated;/inspire-adrauthors them./inspire-task— the task tracker.structurevalidates its on-disk invariants;/inspire-taskoperates the tickets./inspire-module,/inspire-feature— the module/feature reviews this orchestrates and delegates deep dives to./inspire-face— owns faces and bundles and is the diagnostic for their coherence. Review flagsfaces:andbundles:values that don't resolve and screens that bind actions their face does not reach; the fix is that skill's./inspire-process— owns processes; review reportsprocess-shapefindings and routes them there.
04 同仓库更多 Skills
update
Update this project's INSPIRE runtime to the installed plugin's version: report drift, re-materialize unchanged files, and refuse to clobber local edits. Use after /plugin update inspire reports a newer base.
migrate
Move this project's knowledge base from the layered layout (01_adr, 02_modules, 03_features, 04_domain, 05_screens and the surface roster) to the concept layout under inspire_kb/01_spec/, where every artifact has a typed extension, and convert its [[wikilinks]] to relative Markdown links. Run once, on purpose, after /inspire:update reports the layered layout or counts wikilinks. Rebuilds nothing that is already built.
init
Install the INSPIRE runtime into this repository: materialize the skills, validators, hooks and knowledge-base skeleton, register the hooks, and freeze the runtime version. Run once per project. Use when setting up INSPIRE in a new or existing repo.
inspire-task
Task tracker: create / update / close / list / show tickets in inspire_kb/99_tracker/. A plain-file kanban (one Markdown file per ticket, no external tool) for tracking work, drift, and skill-feedback. Use to open, advance, close, or query tickets.
inspire-spike
Vertical spikes: register and harvest external functional-prototype repos. A spike is a narrow, deep, throwaway build in its OWN repo that answers 'can we build it as we think?'. Use to register such a repo (name/reference it, set up its learnings hygiene) and to capture its learnings into the vault as a gap analysis against the current project.
inspire-screens
Create and validate screens (a screen is the UI spec for one view). A screen declares its own bindings — data sources, dispatches, navigation, states — and names a shared layout and shared components as dependencies. Use when designing or reviewing an application's screens.