# Verification Loop

> Use this skill for iterative verify-fix cycles across build, type, lint, test, and security gates until completion or explicit stop conditions.

- Skill: `materialofair/verification-loop` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add materialofair/verification-loop`
- Raw SKILL.md: https://api.skillmd.com/api/skills/materialofair/verification-loop/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: materialofair (https://skillmd.com/u/materialofair)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/materialofair/verification-loop

---



# Verification Loop Skill

> Codex invocation: use `$verification-loop ...` or `verification-loop: ...`

Run repeated verification cycles with bounded retries and clear blocker escalation.

## Capabilities

- Execute ordered quality gates with deterministic stop rules.
- Detect failure class and route to minimal fix scope.
- Repeat verify -> diagnose -> fix until pass or budget exhaustion.
- Produce machine-readable cycle status and blocker details.
- Support pre-PR hardening with security checks.

## Input Requirements

- `goal` (required): full-quality | tests | build | lint | typecheck | pre-pr.
- `scope` (optional): changed files/modules only.
- `max_cycles` (optional): default 5.
- `max_same_failure` (optional): default 3.
- `strict_security` (optional): true | false.

## How to Use

```text
$verification-loop full-quality
$verification-loop pre-pr
$verification-loop tests for src/auth/
```

## Routing Boundary

- Use `$verification-loop` when a single verification pass is not enough and automatic retry/fix cycling is needed.
- Use `$verify` for one-pass verification snapshot.
- Use `$test-gen` for adding missing tests after code changes.

## Native Subagent Protocol (Codex)

Codex supports native subagents. Delegate with `spawn_agent`, coordinate with `send_input`, collect via `wait_agent`, and clean up with `close_agent`.

Minimal orchestration pattern:

```text
spawn_agent -> send_input (optional) -> wait_agent -> close_agent
```

Fallback: execute the loop sequentially when delegation is unavailable.

## Command Resolution

Prefer project-native commands by package manager lockfile detection.

Node defaults:

- build: `npm run build`
- typecheck: `npx tsc --noEmit` (or `npm run typecheck` if defined)
- lint: `npm run lint`
- tests: `npm run test`

Security checks:

```bash
rg -n "sk-[A-Za-z0-9_-]{20,}" --glob '!node_modules' .
rg -n "api[_-]?key|secret|token\s*=\s*['\"]" --glob '!node_modules' .
```

Debug log audit:

```bash
rg -n "console\.log|print\(" src tests
```

## Cycle Algorithm

State file: `.omc/state/verification-loop-state.json`

Per cycle:

1. run selected quality gates in fixed order
2. if all pass -> complete
3. if any fail -> classify failure (build/type/lint/test/security)
4. apply minimal scoped fix
5. re-run failed gate first, then dependent gates
6. record cycle result and repeat

## Stop Conditions

- success: all selected gates pass
- max cycles reached (`max_cycles`)
- same failure repeats `max_same_failure` times
- environment blocker (missing deps/service unavailable)

## Output Contract

```text
[VERIFICATION-LOOP]
- goal
- cycle: <n>/<max_cycles>
- active gate

[RESULT]
- build/type/lint/test/security: pass|fail
- failure class
- fix applied

[NEXT]
- recheck command
- escalation decision

[STATUS]
- result: <complete|blocked|in_progress>
- promise: <PROMISE tag>
- stop_reason: <if blocked>
```

## Completion and Blocking Tags

- `[PROMISE:VERIFICATION_LOOP_COMPLETE]`
- `[PROMISE:VERIFICATION_LOOP_BLOCKED]`

## Continuous Mode Guidance

For long tasks, run this loop at checkpoints:

- before opening PR
- after major refactor
- after dependency upgrades

If loop blocks, handoff to focused skills:

- build/type errors -> `$build-fix`
- test coverage gaps -> `$test-gen`
- security-sensitive failures -> `$security-review`