verify-work

Plugin verify-work from marketplace dev-discipline in mhshaon98/agent-skills. Commands read from the repo's manifest — 2 days ago.

Skills (27)

• antigravity-bridge

  Delegate bounded work or get a third-provider opinion from Google's Antigravity CLI (agy, Gemini). Use on "ask antigravity", "ask gemini", "send this to agy", as a Claude-vs-Codex tie-breaker, or for cheap bounded research sweeps.

• app-audit

  Automated technical and disclosure compliance-assistance audit of a project: profiling, read-only specialist subagents, fresh research, optional Codex peer review, one consolidated report. Use on /app-audit, "audit this app", or a pre-launch security/privacy/store check.

• apply-richformat

  Give a deliverable text surface (document, report, README, app screen, data table, designed CLI output) visible rank and separation; ordinary chat answers are out of scope. Use when building or reviewing one, or on "rich formatting", "text vomit", "needs better structure", "make this readable".

• astra-conductor

  Operating mode for the ENTIRE session whenever the session model is GPT-6-Astra: Astra acts only as architect/orchestrator, delegates ALL execution to gpt-5.6/5.4/5.3 subagents via spawn_agent, and verifies end-to-end.

• bug-ledger

  Use when a bug is reported, found, fixed, or reopened, or before working in a bug-prone area. Check BUG_LIST.md before debugging: many "new" bugs are regressions of fixed ones.

• call-handoff

  Cold-start briefing from the project's newest handoff, open bugs and next-session prompt in one script pass, turns Remote Control on, then stops. Use at session start or on "call handoff", "catch up", "where did we leave off".

• client-cms

  Add a secure client-facing admin/content layer to the current website so nontechnical staff can edit approved content without touching source. Use on /client-cms, "add an admin panel", or when the user wants site content editable by a client or business owner.

• codex-bridge

  Delegate a task to OpenAI Codex or get a non-Claude second review via the codex plugin. Use on "ask codex", "have codex review/check this", "send to codex", or before a release when a second provider is worth it.

• collect-reports

  From the main session, collect a verified status report from every linked, forked and nested Claude Desktop session into one board. Use on "collect reports", "get everyone's status", or "have the forks report back", and whenever a "RECEIPT REQUESTED" message arrives.

• commissioning-logger

  Only when explicitly asked to log a multi-step hands-on process (commissioning, wiring, bring-up, troubleshooting): "log this", "start a commissioning log", "write it up as we go". Keeps log.md plus a Word doc with annotated screenshots.

• compliance-check

  Check what is actually live (store listing and labels, published policies, deployed backend, retention jobs) against the code and published words, then work the remediation runbook. Use on /compliance-check, "what's left on compliance", "close out the audit", or before a store submission.

• cowork-relay

  Split work between Claude Cowork and Claude Code on one project: each side does all it can and ends with a copy-paste relay prompt for the other. Use in a Cowork session or when a RELAY prompt is pasted.

• deepen

  Before planning a substantial build, redesign, feature, research question or subagent brief, answer four questions (what a pro would add, rawest source, their methods, what outside check proves it) and rewrite the brief around them. Deeper, not bigger; the checks become verify-work's done-criteria.

• doc-metadata

  Stamp the project's chosen author and company on every Word, Excel, PowerPoint or PDF file the agent creates or edits, and strip tool and AI fingerprints such as "python-docx" or "generated by". Asks the user once per project which identity to use; also on /doc-metadata or "fix the metadata".

• eng-report

  Apply an engineering house style (vendor-manual layout, the user's own logo and colours, ANSI safety notices, document-control blocks) to a Word, Markdown or PDF document. Use on /eng-report, "make this a proper engineering report", or "format this like a vendor manual"; on first use it asks for the

• fable-conductor

  Session-long operating mode when the session model is Fable/Mythos-class: Fable plans, briefs and verifies, works inline on quick or sequential steps, and delegates gathering, parallel workstreams and large-diff review. Not for sessions whose main model is the frontier executor tier.

• new-project

  Bootstrap a brand-new project with proven architecture defaults and continuity scaffolding. Use when starting a new project from scratch, or when touching a legacy project that has no handoff/git/structure yet.

• pre-release-review

  Independent fresh-context review before anything ships. Use on "ship it", "release", "submit", "deploy", "push to production", before any production push, and after a large diff, even when the diff looks fine.

• report-back

  Close out a forked or side session and send its verified status to the session that started it. Use on "report back", "send your status to the main session", or when collect-reports asks.

• safe-data-write

  Use before any write, migration, or deletion touching user data: stores, schemas, user-owned files (Excel workbooks, databases, documents). Backup, audit and rollback rules learned from shipped data-loss bugs.

• security-pass

  Use when touching auth, API keys, secrets, RLS/policies, payment or personal data, any new endpoint that writes, or live backends — and before any release. Security and secrets checklist; every rule traces to a real production finding.

• session-start

  Old name for the session-start ritual, which now lives in call-handoff; run that skill instead.

• slop-clean

  Project-wide sweep that removes heavy AI slop from design and copy while keeping deliberate, evidenced creative choices (bold colour, density, texture, voice). Manual only (/slop-clean [path]); never self-trigger.

• spawn-agent

  Use BEFORE spawning any subagent (research, implementation, review, debugging) or choosing a model for delegated work. Gate, brief-writing procedure, and model-tier routing; usage limits are a real budget.

• update-handoff

  End-of-session wrap: holistic handoff (this and linked sessions, worktrees), lessons, memory, bug ledger and NEXT-SESSION-PROMPT.md; archives finished nested sessions, Remote Control off. Use on "update handoff", "wrap up", "end the session", or before context runs out.

• usage-here

  Report what this session cost: tokens and dollars by model, share of the 5-hour and weekly limits, top three prompts. Use on "usage here", "what did this session cost", "how much have I burned", "token usage", or questions about limits mid-session.

• verify-work

  Use before claiming work is done, working, or fixed, and before writing a handoff or completion summary. Verification ladder plus the Verified/NOT-verified ledger.

Similar plugins