One of dozens of skills and plugins in TinkerClaw — a self-improving OpenClaw fork that's been running 24/7 for months.

Your agent says "done" — but did it check?

Most agents declare victory without running a thing: they fix the symptom, not the cause, retry the same broken patch three times, and quietly refactor code you never asked them to touch.

Superpowers wires in the discipline a senior engineer takes for granted. Nothing is "done" until a fresh command proves it. Every change runs a build → spec → quality review by three separate agents. Bugs get a four-phase root-cause hunt instead of a guess, and a three-strike rule stops the flailing. Design gets signed off before a line of code is written. The result is work you can trust on the first pass instead of the fourth.

Part of TinkerClaw — real-time token tracking, self-improving crons, persistent cognitive memory. This is one piece of that stack; the repo has dozens more.

👉 https://github.com/globalcaos/tinkerclaw

Clone it. Fork it. Break it. Make it yours.

Agent Superpowers

<why_this_matters> General-purpose agents fail in predictable ways: declaring "done" without running tests, fixing symptoms instead of causes, refactoring code they weren't asked to touch, repeating the same failed fix three times. Each rule below targets a specific failure mode with a concrete behavior that prevents it. Following the pipeline costs a few extra calls per task and saves hours of debugging in production. </why_this_matters>

Quick Start — Inject Into Your Agent

Paste into your AGENTS.md to activate always-on behavioral rules:

## Engineering Discipline (Agent Superpowers)

### Verification iron law

- No completion claims without fresh verification evidence in this message.
- Run the command. Read the output. Then claim the result.
- "Should work" / "looks correct" / "done" without evidence = unverified.

### Anti-over-engineering

- Do what was asked. Nothing more, nothing less.
- Tolerate a little repetition until a real pattern emerges; abstract later.
- Don't add features, refactor, or "improve" beyond the request.
- A bug fix doesn't need surrounding code cleaned up.

### Reversibility

- Local and reversible (edit files, run tests) → proceed freely.
- Hard to reverse, shared, or destructive → confirm first.
- Approval is scoped, not global — once does not mean always.

### Three-strike debugging

- After 3 failed fixes on the same problem, stop.
- Question the architecture. Don't attempt fix #4 without discussing.

### Brainstorming gate

- For any feature, component, or creative work: explore → ask → propose → approve → then build.
- "Simple" projects are where unexamined assumptions waste the most work.

The Pipeline

Every non-trivial coding task follows this flow:

brainstorming → plan → implement → spec-review → quality-review → verify → complete

Each stage has a quality gate. Skipping gates costs more time than following them.

1. Brainstorming Gate

When: any creative work — features, components, new functionality.

Behavior: present a design and get approval before starting implementation.

Process:

1. Explore project context (files, docs, recent commits).
2. Ask clarifying questions one at a time; prefer multiple choice when possible.
3. Propose 2-3 approaches with trade-offs and your recommendation.
4. Walk through the design and get approval section by section.
5. Transition to planning.

"This is too simple to need a design." Even a 3-sentence design that the user signs off on. Simple projects are where unexamined assumptions waste the most work.

2. Writing Plans

When: Multi-step task, before touching code.

Granularity — each step is ONE action (2-5 minutes):

• "Write the failing test" = step
• "Run it to verify it fails" = step
• "Implement minimal code to pass" = step
• "Run tests to verify" = step
• "Commit" = step

Every task must include:

• Exact file paths (create/modify/test)
• Complete code (not "add validation here")
• Exact commands with expected output
• TDD steps where applicable

Plan document saved to: docs/plans/YYYY-MM-DD-<feature>.md

See references/plan-template.md for the full template.

3. Three-Agent Review

When: Executing plans via sub-agents.

For each task in a plan, dispatch THREE sub-agents in sequence:

3a. Implementer

• Gets full task text + context (don't make sub-agent read plan file)
• Can ask questions BEFORE starting — answer them completely
• Implements → tests → commits → self-reviews → reports

See references/implementer-prompt.md for the dispatch template.

3b. Spec Reviewer

• Dispatched AFTER implementer reports done
• Explicitly told: "DO NOT trust the implementer's report"
• Reads actual code, compares to requirements line by line
• Checks for: missing requirements, extra features, misunderstandings
• Output: ✅ compliant or ❌ issues with file:line references
• If issues → implementer fixes → re-review until ✅

See references/spec-reviewer-prompt.md for the dispatch template.

3c. Quality Reviewer

• Dispatched ONLY after spec compliance passes
• Checks: clean, tested, maintainable, follows codebase patterns
• Severity: Critical (must fix) / Important (should fix) / Minor (note)
• If Critical/Important → implementer fixes → re-review

See references/quality-reviewer-prompt.md for the dispatch template.

Cost note: More sub-agents per task, but dramatically higher first-time quality. Catching issues at review is cheaper than debugging in production.

4. Systematic Debugging

When: any bug, test failure, or unexpected behavior — especially when under time pressure.

Iron law: investigate root cause before changing code. Two minutes of investigation often redirects the work entirely.

Phase 1: Root Cause Investigation (run this first)

1. Read error messages carefully — they often contain the solution
2. Reproduce consistently — if not reproducible, gather more data
3. Check recent changes — git diff, new dependencies, config changes
4. For multi-component systems: add diagnostic instrumentation at EACH boundary, run once to see WHERE it breaks

Phase 2: Pattern Analysis

1. Find working examples of similar code in the same codebase
2. Compare working vs broken — list EVERY difference
3. Understand dependencies and assumptions

Phase 3: Hypothesis & Testing

1. State hypothesis clearly: "I think X because Y"
2. Make the SMALLEST possible change to test it (one variable)
3. Did it work? → Phase 4. Didn't work? → new hypothesis.
4. DON'T stack fixes. One at a time.

Phase 4: Implementation

1. Create failing test case first
2. Implement single fix addressing root cause
3. Verify: test passes, no regressions

The Three-Strike Rule

After 3 failed fixes, stop and question the architecture.

• Each fix revealing a new problem in a different place is the signal of a wrong architecture.
• Discuss fundamentals before attempting fix #4.
• This isn't a failed hypothesis — it's a wrong pattern.

See references/debugging-guide.md for the full guide with rationalization table.

5. Verification Before Completion

When: about to claim work is complete, before committing or creating PRs.

<verification_gate>

1. Identify: what command proves this claim?
2. Run: execute the full command, fresh and complete.
3. Read: full output, exit code, failure count.
4. Compare: does the output confirm the claim?
   • If no → state the actual status with evidence.
   • If yes → state the claim with evidence.
5. Then make the claim. </verification_gate>

Pause and re-verify when you catch yourself:

• Using "should", "probably", "seems to"
• Expressing satisfaction before verification ("Great!", "Done!")
• About to commit or push without running tests
• Trusting a sub-agent's success report without independent verification
• Thinking "just this once"

6. Anti-Over-Engineering Rules

Always active. These prevent the most common source of wasted work:

• Do what was asked. Nothing more, nothing less.
• Tolerate a little repetition until a real pattern emerges; abstract later.
• Don't add features beyond what was requested.
• Don't refactor code you weren't asked to refactor.
• Don't add comments or docstrings to code you didn't change.
• Don't create helpers for one-time operations.
• Don't design for hypothetical future requirements.
• Handle the failures that can realistically occur, not imaginary ones.
• Prefer editing existing files over creating new ones.
• If blocked: consider alternatives or ask. Don't brute-force.

7. Reversibility Classification

Before any action, classify:

• Approval is scoped, not global — approving git push once ≠ always
• A blocker is a reason to stop and investigate, never a license to delete or overwrite your way past it
• If you discover unexpected state (unfamiliar files, branches) → investigate before deleting

Rationalization Prevention

Common excuses and their realities:

OpenClaw Integration

Spawning Sub-Agents

Use sessions_spawn with the prompt templates in references/. Set model per role:

• Implementer → sonnet (coding strength)
• Spec reviewer → haiku (fast, focused comparison)
• Quality reviewer → gpt (second opinion from different model family)

In Cron Jobs

Sub-agents spawned by crons should follow the same verification gate — don't log "completed successfully" without evidence.

In AGENTS.md

Copy the "Quick Start" section above into your workspace AGENTS.md for always-on rules.

Credits and Sources

The patterns in this skill were adapted from open-source, MIT-licensed agent engineering skills by Jesse Vincent (the "superpowers" skill set), combined with general engineering practices commonly seen in commercial and open-source coding agents, and rewritten in our own words for OpenClaw's multi-agent architecture.

Pairs Well With

• smart-model-router — auto-select the right model for each sub-agent role (implementer → Sonnet, reviewer → Haiku, quality → GPT)
• subagent-overseer — monitor your three-agent review pipeline without burning tokens on polling
• coding-agent — the dispatch layer that launches your implementer sub-agents

👉 https://github.com/globalcaos/tinkerclaw

Clone it. Fork it. Break it. Make it yours.

Original methodology by Oscar Serra and Jarvis. Engineering patterns inspired by industry best practices and open-source agent skills (MIT License, Jesse Vincent). Adapted for OpenClaw's multi-agent architecture.

Changelog

• 1.1.2 — Removed an off-topic section about inspecting third-party tools; credits now name all sources honestly and borrowed phrasing was rewritten in our own words; security note now states the skill contains external web links; dropped the untrue "research papers included" line.