Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add oleg494/coding-kit --skill superpowersgit clone --depth 1 https://github.com/oleg494/coding-kitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/oleg494/coding-kit/superpowers)<a href="https://agentmods.dev/skills/oleg494/coding-kit/superpowers"><img src="https://agentmods.dev/badge/skills/oleg494/coding-kit/superpowers/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/oleg494/coding-kit/superpowers"><img src="https://agentmods.dev/badge/skills/oleg494/coding-kit/superpowers.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00050 | $0.01198 |
| Opus 5.5 | $0.00020 | $0.00479 |
| Sonnet 5.5 | $0.00010 | $0.00240 |
| Haiku 4.5 | $0.00005 | $0.00120 |
Grade A, and why
superpowers scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Superpowers — main development method
The core method lives in OPS §§2–3; load this skill when phase detail resolves an open question. Scale the cycle to the task's scope and risk.
The Cycle
PLAN ──→ TDD ──→ IMPLEMENT ──→ VERIFY ──→ REPORT
│ │ │ │ │
▼ ▼ ▼ ▼ ▼
Spec Red test Green code Evidence Outcome
first first minimal observed first
Phase skills (obra accretion)
Kit v2: each phase has a granular skill helper. A phase is not replaced, but deepened:
- PLAN →
brainstorming(design questions, spec),writing-plans(execution plan) - IMPLEMENT →
dispatching-parallel-agents(parallel slices), per-plan implementation with checkpoints - VERIFY →
verification-before-completion(applicable evidence for the current state),requesting-code-review,fable-judge(adversarial) - Debug →
systematic-debugging - Git →
using-git-worktrees,finishing-a-development-branch
These are pointers, not a load order. A helper loads for an unresolved question in its domain (a real design question, a verification-honesty problem, a debugging unknown) — never merely because a phase named it, and never cascaded: loading one helper does not pull the others. The host's mandatory skill policy always wins over this thrift.
Phase 1: PLAN
Formulate what "done" means — concrete, observable.
- What should be true when the task is done?
- Which files do you touch? Which do you NOT touch?
- What assumptions do you make?
- Split by independently verifiable outcomes and shared interfaces, not file or step counts.
Scope discipline: touch only what the task requires. Not "I'll clean up along the way".
Phase 2: TDD
Red test → green code → refactoring.
- Define an observable check before changing behavior; for a bug, reproduce the failure where reproduction is needed for diagnosis or regression defense — a user-reported failure is ground truth for the symptom, and prior failure evidence valid for the unchanged state counts.
- Test names express consumer rules, not implementation details. Keep a permanent regression only where a plausible bug would fail it; a throwaway probe or rendered UI interaction can provide the appropriate proof.
- A test verifies behavior, not implementation.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago Changed · +2 lines b9edbb2896bc
- 8d ago Changed · +4 tokens per session b01d90b34775
- 11d ago Changed · +6 lines c3f93eced706
- 25d ago Changed 6e356546d156
- 26d ago Changed a6728e03e552
- 1mo ago Changed de4df6db8d79
- 1mo ago Changed · -1 lines · -24 tokens per session 9d1295e3ccbb
- 1mo ago Changed 5316328e5db9
- 1mo ago Changed 4cbf15d5300a
- 1mo ago First seen · 115 lines · 70 tokens per session scan A 260a952a33cb
superpowers is a skill published in the GitHub repository oleg494/coding-kit (3 stars, last pushed 2d ago), licensed MIT. It adds 50 tokens to every session and 1,198 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
writing-skills
Use when creating a new skill, editing an existing skill, or verifying skills work before deployment.
repro-first
Bug-fix mode that reproduces a bug in a failing test before changing any code. Entered with /repro-first.
tdd-implement
Implements a feature or fix test-first: write failing tests, make them pass with the least code, then refactor. Tests track acceptance criteria, not implementation details. Trigger on: "implement with TDD", "write the tests first", "TDD workflow", "red-green-refactor".
tdd-dev-cycle
A test-driven development workflow, meaning developers write automated tests for expected behavior before implementing the code. It defines stages for analysis, testing, development, verification, commits, and repeated fixes.
test-driven-development
Use when implementing new features or fixing bugs. Write tests before implementation following Red-Green-Refactor. Covers JUnit5, MockK, Compose test rules, and the Prove-It pattern for bugs.
python-testing
Python testing strategies using pytest, TDD methodology, fixtures, mocking, parametrization, and coverage requirements.