verify-interpreter
Interpret the output of `specgraph verify` and explain what each status means and what to do next.
- License 许可
- MIT
- 最近更新
- 2026-06-23
- Stars
- ★ 15
- Forks
- 1
01 前言 Front Matter
- name
- verify-interpreter
- description
- Interpret the output of `specgraph verify` and explain what each status means and what to do next.
03 SKILL.md 正文
Read and explain the output of specgraph verify in plain language, then recommend next steps.
Status meanings
| Status | Meaning | Action |
|---|---|---|
PASS |
All required evidence thresholds met | Nothing required |
WARN |
Evidence exists but below required strength for this state | Strengthen evidence or lower requirement |
FAIL |
No evidence found for a required dimension | Add evidence or add waiver |
INSUFFICIENT |
Evidence found but strength is below required level | Raise evidence strength (e.g. E0 → E2) |
SKIP |
Spec has no evidence requirements (draft/deprecated) | Review if state is correct |
WAIVED |
Requirement waived — check expiry | Verify waiver is still valid |
Example output
PASS FEAT-001 [accepted]
WARN FEAT-002 [in_progress]
warn: No verification claims found (advisory)
FAIL FEAT-003 [accepted]
fail: No implementation claims found. Required: E2 (indexed)
INSUFFICIENT FEAT-004 [accepted]
insufficient-evidence: implementation: found E0 but required E2
Verify commands reference
specgraph verify # summary table for all specs
specgraph verify --spec FEAT-001 # single spec only
specgraph verify --json # machine-readable output
specgraph verify --changed src/foo.ts # incremental scan (changed files only)
specgraph verify --no-db # dry run — skip writing to DB
specgraph explain <SPEC-ID> # full claim breakdown for one spec
specgraph explain # all specs
specgraph find --spec FEAT-001 # locate files that reference a spec
specgraph find --relation IMPLEMENTS # all IMPLEMENTS claims
specgraph find --provider annotation # claims from a specific provider
specgraph find --strength 1 # E1 and above
specgraph waivers # list active waivers
specgraph waivers FEAT-001 # waivers for one spec
specgraph subject issue:beads:bd-a1b2 # show a specific Beads issue's claims
specgraph subject symbol:annotation:MyClass # show a symbol's claims
Subject IDs follow the format kind:provider:identity (e.g. issue:beads:bd-a1b2, symbol:annotation:TodoStore).
Interpretation checklist
- Count PASS / WARN / FAIL / INSUFFICIENT counts — overall health indicator.
- For each FAIL: run
specgraph explain <ID>to see what claims exist. - For each INSUFFICIENT: check the
required_evidencethreshold in the spec frontmatter — lower it or add stronger evidence. - For each WARN: check if the evidence level matches the spec's
staterequirements. - Check waiver expiry dates — expired waivers flip to WARN/FAIL.
- Confirm provider scans are recent — stale scan data can hide real gaps.
Common causes
| Symptom | Likely cause | Fix |
|---|---|---|
| All specs FAIL | Annotations not scanned | Run specgraph verify again; check @spec syntax |
INSUFFICIENT on accepted specs |
Evidence is E0/E1, need E2+ | Add cross-refs or tests to raise strength |
WARN on accepted specs |
Evidence is E0/E1, need E2+ | Add cross-refs or tests |
| Unexpected PASS | Waiver is covering a gap | Check specgraph waivers for active waivers |
| Missing specs in output | Spec file not in docs/ |
Check path and .specgraph/config.json |
| Beads claims missing | Beads not enabled | Set beads.enabled: true in .specgraph/config.json |
Rules
- Never dismiss a FAIL without understanding the root cause.
- Use
specgraph explainbefore deciding whether to add evidence or a waiver. - Report the exact spec IDs and dimensions in any gap summary you produce.
- Config file is
.specgraph/config.json(falls back to.agent-docs/config.json).
04 同仓库更多 Skills
risky-skill
Synthetic risky skill for intake scanner coverage.
fixture-skill
Synthetic skill for intake scanner coverage.
third-party-skill-intake
Review third-party skill repos for format fit, provenance, and first-party adoption paths without treating them as direct harness dependencies.
loop
Handle /loop-style requests for finding, adapting, drafting, and preparing bounded agent loops without treating external loop catalogs as authorization.
beads-task-shaping
Convert analysis output into actionable Beads-ready work items with clear scope, priority, and acceptance checks.
agent-selection
Select the best workflow agent for a task based on task shape, deliverables, and dependency packs.