Metadata
| Status | done |
|---|---|
| Assigned | agent-1209 |
| Agent identity | f51439356729d112a6c404803d88015d5b44832c6c584c62b96732b63c2b0c7e |
| Model | claude:opus |
| Created | 2026-04-30T02:05:27.749255763+00:00 |
| Started | 2026-04-30T02:14:13.044071680+00:00 |
| Completed | 2026-04-30T02:17:20.694958742+00:00 |
| Tags | fix,docs,audit,sweep, eval-scheduled |
| Eval score | 0.93 |
| └ blocking impact | 0.90 |
| └ completeness | 0.95 |
| └ coordination overhead | 0.95 |
| └ correctness | 0.95 |
| └ downstream usability | 0.95 |
| └ efficiency | 0.85 |
| └ intent fidelity | 0.82 |
| └ style adherence | 0.95 |
Description
Description
Run wg --help-all (lists every command). For each command, grep across docs/ + README.md + CLAUDE.md + SKILL.md to find at least one reference. List any commands that appear in --help but have ZERO documentation. Coverage check, not deep-dive.
Output
Post findings via `wg log `. Structured list of items found. NO file modifications — audit only. Synthesis task applies fixes.
Validation
- Findings posted via wg log
- Specific file:line citations where possible
- No source / doc modifications
Depends on
Required by
Log
- 2026-04-30T02:05:27.721129871+00:00 Task paused
- 2026-04-30T02:10:06.465374629+00:00 Task published
- 2026-04-30T02:14:13.044080546+00:00 Spawned by coordinator --executor claude --model opus
- 2026-04-30T02:14:21.634287443+00:00 Starting audit: extracting full command list from wg --help-all then grepping docs for each
- 2026-04-30T02:17:10.473411758+00:00 AUDIT FINDINGS: wg --help-all coverage in user-facing docs Scope: 120 commands (output of `wg --help-all`) Doc surface searched: docs/ + README.md + CLAUDE.md + .claude/skills/wg/SKILL.md Living vs archive: docs/archive/ excluded for "living coverage" verdict Search regex: `\bwg <cmd>\b` and backticked `wg <cmd>` forms Sanity check: list=81 hits, show=169, add=812, done=523 (regex works) === ZERO-COVERAGE COMMANDS (10) === None of these appear in any living doc OR archive doc as `wg <cmd>`: 1. claude-handler — Bridge Claude CLI stream-json stdio ↔ chat/<ref>/*.jsonl (likely intentional: internal subprocess bridge, not user-facing) 2. codex-handler — Bridge Codex CLI JSONL output ↔ chat/<ref>/*.jsonl (likely intentional: internal subprocess bridge) 3. executors — List executors wg knows about, where binaries live (FIX RECOMMENDED: useful diagnostic; only ref is task-name match in docs/research/amplifier-executor-gap.md) 4. html — Render workgraph as static, clickable HTML viewer (TUI-parity) (FIX RECOMMENDED: user-facing feature, deserves doc mention) 5. insert — Insert task at position relative to existing target (Help text says "foundation for `wg rescue`"; rescue is also undocumented: grep -rE "\bwg (insert|rescue)\b" docs/ README.md CLAUDE.md → 0 hits) 6. reprioritize — Change task's priority level (critical, high, normal, low, idle) (FIX RECOMMENDED: workflow-relevant, missing from COMMANDS.md and AGENT-GUIDE.md) 7. secret — Manage secrets (API keys) in credential store (FIX RECOMMENDED: docs/SECURITY.md and docs/KEY_DOCS.md mention "secret" the word but never `wg secret` invocation; setup wizards reference it implicitly) 8. tui-nex — Interactive agentic TUI (ratatui-based nex) (Possibly experimental/undocumented-on-purpose) 9. tui-pty — PTY-embedded nex TUI (Possibly experimental/undocumented-on-purpose) 10. which — Print resolved workgraph directory + resolver step (FIX RECOMMENDED: ESSENTIAL debug command, exact help text says "useful when you're confused about which graph `wg add` is talking to" — agents hit this confusion regularly) === LOW-COVERAGE COMMANDS (worth noting) === - reap — only in docs/research/agent-lifecycle-and-kill-mechanics.md:264-265 (proposed-as-future-design, predates implementation; no current doc) - incomplete — only in docs/design-actor-driven-cleanup.md:92,99 (referenced as state transition; missing from agent-facing docs) - rescue — 0 hits in living docs; 0 in archive (parallel gap with `insert`; both are graph-surgery primitives) === METHODOLOGY NOTES === - 120/120 commands extracted via `wg --help-all` parsing - Per-command grep counted both `wg <cmd>` and backticked forms - Archive count subtracted to compute "living coverage" - All 10 zero-coverage commands re-verified individually - 110 commands have at least one living-doc reference (~92% coverage) === SUGGESTED ACTIONS FOR doc-sync-audit (synthesis task) === Priority 1 (user-facing utilities the user will need): - Add `wg which` to docs/COMMANDS.md (debugging section) and CLAUDE.md - Add `wg secret` to docs/SECURITY.md (currently lists keys but not the cmd) - Add `wg html` to docs/COMMANDS.md (visualization section) - Add `wg reprioritize` to docs/COMMANDS.md and docs/AGENT-GUIDE.md Priority 2 (graph-surgery primitives): - Add `wg insert` and `wg rescue` to docs/COMMANDS.md (advanced section) - Add `wg executors` to docs/COMMANDS.md (config section) Priority 3 (decide whether to document or hide): - tui-nex, tui-pty: confirm whether ready for users - claude-handler, codex-handler: likely keep undocumented (subprocess bridges) NO FILES MODIFIED (audit-only task per description).
- 2026-04-30T02:17:18.397089803+00:00 Validated: findings posted via wg log with file:line citations; git status clean (no source/doc modifications); validation criteria met
- 2026-04-30T02:17:20.694966226+00:00 Task pending eval (agent reported done; awaiting `.evaluate-*` to score)
- 2026-04-30T02:21:16.311029465+00:00 PendingEval → Done (evaluator passed; downstream unblocks)