# Verify

> Lightweight verification gate — prove a claim before making it. Use when finishing ad-hoc work, one-off fixes, or any time you're about to say "done" or "fixed" outside of /execute-phase. Trigger on "verify this works", "prove it", "check your work", "does this actually work". Also referenced inline by /diagnose (Phase 4) and /execute-phase (completion criteria) as an embedded methodology.

- **Type:** Skill
- **Install:** `agentstack add skill-takleb3rry-claude-global-appbuilding-skills-verify`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [takleb3rry](https://agentstack.voostack.com/s/takleb3rry)
- **Installs:** 0
- **Category:** [AI & ML](https://agentstack.voostack.com/c/ai-and-ml)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [takleb3rry](https://github.com/takleb3rry)
- **Source:** https://github.com/takleb3rry/claude-global-appbuilding-skills/tree/main/skills/verify

## Install

```sh
agentstack add skill-takleb3rry-claude-global-appbuilding-skills-verify
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Verify

Prove it works. Then say it works. Never the other way around.

---

## The Checklist

### 1. IDENTIFY
What command(s) would prove the claim you're about to make? Name them explicitly.
- Test suite: `npm test`, `pytest`, `vitest run [file]`
- Build: `npx tsc --noEmit`, `npm run build`
- Specific behavior: `curl`, browser check, database query
- If you can't name a command that proves it, you can't claim it.

### 2. RUN
Execute the command(s). No dry runs. No "I would run." Actually run it.

### 3. READ
Read the full output. Don't skim. Don't stop at the first green line. Look for:
- Partial failures buried in passing output
- Warnings that indicate the test isn't testing what you think
- Skipped tests that cover the relevant behavior

### 4. CHECK
Does the output confirm the claim?
- **Yes:** State the claim with the evidence. "Fixed — `vitest run src/auth` passes 12/12 tests including the new regression test."
- **No:** Do not restate the claim with caveats. Investigate. Return to the work.

---

## Language Rules

| Instead of... | Use... |
|---|---|
| "Should work now" | [run the command, show the output] |
| "I believe this resolves it" | [run the test, cite the result] |
| "That should be fine" | [verify it, state what you verified] |
| "This probably fixes it" | [prove it fixes it] |
| "Done!" (before evidence) | [show evidence, then state completion] |

If you can't point to command output that proves the claim, the claim isn't ready.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [takleb3rry](https://github.com/takleb3rry)
- **Source:** [takleb3rry/claude-global-appbuilding-skills](https://github.com/takleb3rry/claude-global-appbuilding-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** yes
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-takleb3rry-claude-global-appbuilding-skills-verify
- Seller: https://agentstack.voostack.com/s/takleb3rry
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.