v1.0.0 to v1.0.0

3 added, 22 removed. Audit A to A.

---
name: verification-loop
display-name: Verification Loop
description: 6-stage automated verification pipeline producing READY/NOT READY verdict
version: 1.0.0
category: quality
user-invocable: true
- tags:
- [
- verification,
- build,
- typecheck,
- lint,
- test,
- security,
- diff-review,
- ci,
- quality,
- ]
- capabilities:
- [
- build-verification,
- type-checking,
- lint-checking,
- test-execution,
- security-scanning,
- diff-review,
- ]
+ tags: [verification, build, typecheck, lint, test, security, diff-review, ci, quality]
+ capabilities: [build-verification, type-checking, lint-checking, test-execution, security-scanning, diff-review]
tools:
- command_executor
- file_reader
- code_analyzer
instructions: |
Use this skill when the user wants to verify code readiness before commit, merge, or deploy.
Runs a 6-stage verification pipeline: Build → TypeCheck → Lint → Test → Security → DiffReview.
Each stage produces a PASS/FAIL result. The final verdict is READY (all pass) or NOT READY
(with failure details). Automatically detects project type (Node.js, Python, Java) and adapts
commands accordingly. Can skip stages with --skip flag.
examples:
- input: "/verification-loop"
output: "6/6 stages passed. Verdict: READY"
- input: "/verification-loop src/ --skip typecheck"
output: "5/5 stages passed (typecheck skipped). Verdict: READY"
- input: "/verification-loop --stages build,test,security"
output: "3/3 stages passed. Verdict: READY"
model-hints: sonnet
os: [win32, darwin, linux]
author: ChainlessChain
- execution-capabilities: [data:result, data:task, filesystem:read, host:logger, host:skill, process:cwd, process:execute, runtime:time]
+ execution-capabilities: [data:result, data:task, filesystem:read, host:logger, host:process, host:skill, process:cwd, process:execute, runtime:time, system:inspect]
handler: ./handler.js
---
# Verification Loop
## Description
6-stage automated verification pipeline that produces a READY/NOT READY verdict. Inspired by the everything-claude-code verification loop pattern.
## Usage
```
/verification-loop [path] [options]
```
## Options
- `--skip <stages>` - Comma-separated stages to skip (e.g. `--skip typecheck,security`)
- `--stages <stages>` - Run only specified stages (e.g. `--stages build,test`)
- `--verbose` - Show full output from each stage
## Verification Stages
| Stage | Description | Tool |
| ---------- | --------------------------------- | ------------------- |
| Build | Compile/bundle the project | npm run build / mvn |
| TypeCheck | Run static type checking | tsc --noEmit |
| Lint | Run linter for code style issues | eslint / flake8 |
| Test | Execute test suite | vitest / jest |
| Security | Scan for secrets and OWASP issues | security-audit |
| DiffReview | AI review of uncommitted git diff | LLM analysis |
## Execution Flow
```
1. Detect project type (Node.js / Python / Java)
2. Build stage → PASS/FAIL
3. TypeCheck stage → PASS/FAIL
4. Lint stage → PASS/FAIL
5. Test stage → PASS/FAIL
6. Security stage → PASS/FAIL
7. DiffReview stage → PASS/FAIL
8. Aggregate results → READY / NOT READY
```
## Output Format
```
Verification Loop Report
========================
Project: desktop-app-vue (Node.js)
| Stage | Status | Duration | Details |
| ---------- | ------ | -------- | ---------------- |
| Build | PASS | 12.3s | Clean build |
| TypeCheck | PASS | 4.1s | 0 errors |
| Lint | PASS | 2.8s | 0 warnings |
| Test | PASS | 8.5s | 157/157 passed |
| Security | PASS | 1.2s | 0 findings |
| DiffReview | PASS | 3.0s | No issues found |
Verdict: READY
```