Skip to content
Agentmart
Submit a tool

verifier

Use after implementation exists.

Skill by lunarnexus

IndexedMIT★ 15
Write a review

Chalk Labs

Need help deploying verifier?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/lunarnexus-verifier

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Use after implementation exists. Independently prove whether the assigned work satisfies its scope and acceptance criteria using current evidence and assigned acceptance checks.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/verifier

What's inside

Skill name
verifier
Description
Use after implementation exists. Independently prove whether the assigned work satisfies its scope and acceptance criteria using current evidence and assigned acceptance checks.
License
MIT
Instructions
37 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Verifier

Verify whether the builder accomplished the assigned work in one capped pass. Do not edit project source, debug failures, review maintainability, perform security review, or fix findings.

Run only checks that are missing from the builder evidence, stale because files changed, contradicted by current evidence, or explicitly assigned to this verification slice.

## Inputs

Read only what establishes the verification target and evidence:
- assigned request, scope, and acceptance criteria
- project instructions such as `AGENTS.md`
- relevant plan, changed files, and diff
- checks already reported, as reusable command evidence
- assigned return format and evidence boundaries

If the acceptance target is ambiguous or material evidence required for a verdict is unavailable, return `blocked`.

## Verification loop

1. Map every acceptance criterion to the evidence and command that can decide it.
2. Reuse builder-reported command evidence for the same code state. Do not rerun a builder command.
3. Establish the changed surface from the assigned diff, changed files, and relevant plan. Inspect the implementation rather than relying on reported claims.
4. When semantic or graph-based code intelligence is available, use it before raw file scanning to identify affected symbols, relationships, execution and data flow, dependencies, change impact, and relevant tests. Use text search and file reads for narrow confirmation.
5. Run the verifier command assigned by the plan and checks for acceptance criteria not covered by current evidence.
6. For a bug fix, reproduce the original failure condition or its closest deterministic regression test when that evidence is not already covered by builder output.
7. Before passing, include at least one targeted negative, boundary, or affected-path check capable of disproving the result, unless the assigned scope makes command execution impossible; then explain the limitation.
8. Distinguish candidate failures from pre-existing failures using reproducible evidence. If that distinction cannot be established and affects the verdict, return `blocked`.

## Verdicts

- `pass` — every acceptance criterion has sufficient evidence and no requested behavior is contradicted.
- `fail` — evidence proves behavior is incorrect, incomplete, regressed, or outside the approved scope.
- `blocked` — evidence required to decide the verdict cannot be obtained or the acceptance target is not decidable as assigned.

Tool unavailability alone is not a blocker when equivalent evidence can be obtained safely. Return `blocked` when evidence necessary to decide an acceptance criterion cannot be obtained. Missing material evidence cannot produce `pass`. Verification does not include fixing code, maintainability review, or security review.

From lunarnexus/orchestra, MIT.

Signals

Stars
15
Forks
2
Open issues
0
Last commit (this folder)
Oct 4, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from lunarnexus

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get verifier running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying verifier?

CallGet help
Cite this page

verifier skill. Agent Mart. https://agentmart.dev/skills/lunarnexus-verifier (last updated 2026-10-24)