claude-damn
A SME of SWE: a plugin giving Claude Code the expertise of an Expert Software Engineer across every SWE area, while still inviting novice pr
Open source Open in the app JSON README (API)
About
A SME of SWE: a plugin giving Claude Code the expertise of an Expert Software Engineer across every SWE area, while still inviting novice programmers to make gains on best-practice rails. Layers on top of `obra/superpowers`. Ships: - 38+ skills in super/duper/cat/tdd modifier families (20+ combinatoric variants) - Multi-phase `/expert-review` with language-specific + cloud + security lenses - `/sme-test` coaching-driven test authoring - `/tesseract` cross-session memory via file I/O (anchor + shelf + bulk-beings) - `/checkpoint-save` + `/checkpoint-resume` for work-state across sessions - Pre-tool `block-inline-scripts` hook (blocks multiline non-Bash scripts + enforces command length / separator limits) - Per-session cost tracking (`/cost_`, `/cost-opt`) via JSONL logs - Opinionated `CLAUDE.md` with model routing, Python/FastAPI style, pytest/behave standards Post-install, skills are invoked via slash commands: `/tdd`, `/super-duper-cat`, `/expert-review`, `/checkpoint-save`, `/tesser
Details
- Kind
- Plugins
- Topic
- AI, RAG & memory
- Publisher
- laurac8r
- Origin
- marketplace
- Category
- ferramentas
- Stars
- 3
- Forks
- 1
- Last push
- 2026-08-15T12:26:07Z
- Repository state
- ativo
- Language
- Python
- License
- MIT
- Added
- 2026-08-30 01:48:58
- Updated
- 2026-08-30 01:48:58
- Origin id
laurac8r/claude-damn/claude-damn
README
# claude-damn
A composable skill library and opinionated harness for
[Claude Code](https://docs.anthropic.com/en/docs/claude-code). Ships slash
commands, skills, hooks, permission guards, and cost-tracking tooling that layer
on top of the
[superpowers](https://github.com/anthropics/claude-plugins-official/tree/main/superpowers)
plugin.
<img src="img/home-meme.png" alt="home meme" width="800">
> **✅ Status:** `claude-damn` is **live** in the
> [`anthropics/claude-plugins-community`](https://github.com/anthropics/claude-plugins-community)
> marketplace. The published marketplace version is **v1.12.0**; the working
> tree is at **v1.12.2**, pending the next batch sync. See `CHANGELOG.md` for
> what's in each version.
Install:
```shell
claude plugin marketplace add anthropics/claude-plugins-community
claude plugin install claude-damn@claude-community
```
See [`ROADMAP.md`](ROADMAP.md) for what's next and
[`CHANGELOG.md`](CHANGELOG.md) for what's landed. Design specs live in
`docs/superpowers/{specs,plans}/`.
## What's in the box
### Skills (slash commands)
Skills compose from four modifiers — stack them to build the workflow you need:
| Modifier | What it adds |
| -------- | --------------------------------------- |
| `super` | Brainstorm requirements first, then TDD |
| `duper` | Isolated git worktree |
| `cat` | Parallel subagent execution |
| `tdd` | Test-driven development |
These combine into four skill families:
- **TDD** — `/tdd`, `/tdd-cat`, `/duper-tdd`, `/duper-tdd-cat`
- **Brainstorm + TDD** — `/super`, `/super-cat`, `/super-duper`,
`/super-duper-cat`
- **Debug + Fix (fixer)** — `/fixer`, `/super-fixer`, `/super-fixer-cat`,
`/super-duper-fixer`, `/super-duper-fixer-cat`, `/super-duper-fixer-tdd-cat`
- **Expert Review** — `/expert-review` through
`/expert-super-duper-tdd-cat-review` (14 combinatoric variants), plus
`/expert-final-review` (pre-merge go/no-go gate)
Plus utility skills: `/atlas`, `/checkpoint-save`, `/checkpoint-resume`,
`/check-yourself`, `/cost_`, `/cost-opt`, `/fast-pr-feedback-to-me`,
`/fast-pr-feedback-to-others`, `/fast-pr-final-self-review`, `/learn`,
`/legal-visual-aid`, `/listen`, `/pause`, `/proceed`, `/review`, `/sme-test`,
`/sync`, `/task-list`, `/tesseract`, `/visual-aid`, and more.
See [`skills/README.md`](skills/README.md) for the full combinatoric table.
### Harness features
- **Permission guards** — granular allow/ask/deny rules in `settings.json`
(blocks destructive GitHub CLI operations, gates commits, protects settings
files)
- **Inline-script hook** — pre-tool hook that blocks multiline non-Bash scripts
from running inside the Bash tool
- **Cost tracking** — `/cost_` extracts per-session token usage from JSONL logs;
`/cost-opt` compacts and reviews optimization suggestions
- **Checkpoint system** — save/resume work across sessions with automatic
archive rotation per branch
- **CLAUDE.md** — opinionated project rules for model routing, testing
standards, Python style, and FastAPI architecture
## Setup
### Prerequisites
- [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code) installed
and authenticated
- Git
### Install from the marketplace
`claude-damn` is published in the
[`anthropics/claude-plugins-community`](https://github.com/anthropics/claude-plugins-community)
marketplace. If you haven't already added that marketplace, do so first:
```shell
claude plugin marketplace add anthropics/claude-plugins-community
```
Then install:
```shell
claude plugin install claude-damn@claude-community
```
This will register every skill, command, and the inline-script PreToolUse hook.
Restart or reload the session once to pick up the hook.
To pull the latest published version later:
```shell
claude plugin update claude-damn@claude-community
```
### Companion plugins
`claude-damn` layers on top of `superpowers`. Install it if you don't have it
already:
```shell
/plugin install superpowers@claude-plugins-official
```
Other marketplace plugins the skill set composes with (all optional):
| Plugin | What it adds |
| --------------------- | ---------------------------- |
| `code-review` | PR review workflows |
| `frontend-design` | UI/UX skill |
| `code-simplifier` | Code cleanup |
| `feature-dev` | Guided feature development |
| `skill-creator` | Create/edit skills |
| `security-guidance` | Security audit |
| `pr-review-toolkit` | Multi-agent PR review |
| `agent-sdk-dev` | Agent SDK scaffolding |
| `hookify` | Hook rule creation |
| `chrome-devtools-mcp` | Browser DevTools via MCP |
| `playground` | Interactive HTML playgrounds |
| `remember` | Session handoff notes |
### Verify
Open a new Claude Code session and invoke any skill — e.g.:
```
/expert-review
```
The skill content should expand in your session.
### Personalize (optional)
Operator-specific preferences (commit style, model routing, subagent delegation
targets) go in `~/.claude/rules/PERSONALIZATION.md`. Copy the template and edit
to taste:
```shell
cp rules/PERSONALIZATION.example.md ~/.claude/rules/PERSONALIZATION.md
```
`CLAUDE.md` holds general engineering rules that apply to every clone and should
stay in sync with upstream.
### Manual install (for contributors)
If you're hacking on `claude-damn` itself, install from a local clone:
```shell
git clone https://github.com/laurac8r/claude-damn
cd claude-damn
uv sync # dev/test dependencies
uv run pytest # 719 tests green (777 total; smoke + performance deselected by default)
```
Then point Claude Code at the local checkout so skills, commands, and the
PreToolUse hook load from your working tree (not the published marketplace
copy):
```shell
claude --plugin-dir /absolute/path/to/claude-damn
```
Or use `--add-dir` to allow tool access without registering as a plugin:
```shell
claude --add-dir /absolute/path/to/claude-damn
```
When you're done hacking, drop the local install and go back to the marketplace
copy:
```shell
claude plugin install claude-damn@claude-community
```
## Quickstart
A few skill invocations to get a feel for the workflow:
```claude-like
/tdd # Start a test-driven development loop
/super # Brainstorm first, then TDD
/super-duper-cat # Brainstorm + worktree isolation + parallel subagents
/expert-review # Multi-phase expert code review
/checkpoint-save # Save resumable work state
/checkpoint-resume # Pick up where you left off
/cost_ # Log this session's token spend
```
See [`skills/README.md`](skills/README.md) for the full combinatoric table of
how `super` / `duper` / `cat` / `tdd` compose into 20+ skill variants.
## Project structure
```tree-like
.
├── .claude-plugin/
│ └── plugin.json # Plugin manifest (name, version, author, keywords)
├── CLAUDE.md # Project instructions (model routing, style, testing)
├── README.md # This file
├── LICENSE # MIT
├── settings.json # Default harness config (permissions, hooks, plugins)
├── pyproject.toml # Python tooling (uv, pytest, ruff, ty)
├── skills/ # Auto-discovered skills (one SKILL.md per skill)
│ ├── README.md # Full combinatoric skill table
│ ├── tdd/ # Base TDD workflow
│ ├── super/ # Brainstorm + TDD
│ ├── duper/ # Git worktree isolation
│ ├── cat/ # Subagent-driven development
│ ├── expert-review/ # Multi-phase expert review
│ ├── expert-*-review/ # 13 combinatoric variants + /expert-final-review gate
│ ├── checkpoint-save/ # Save resumable checkpoint
│ ├── checkpoint-resume/ # Resume from checkpoint
│ ├── tesseract/ # Cross-session memory via file I/O
│ └── ...
├── hooks/
│ ├── hooks.json # PreToolUse registration
│ └── block-inline-scripts.py # Guardrail for multiline non-Bash scripts
├── src/
│ └── extract_cost.py # Token-usage parser for session JSONL logs
├── scripts/
│ └── statusline-command.sh # Per-session cost in the statusline
├── tests/ # pytest + bats suite
└── docs/
└── superpowers/ # Design specs (sme-test, sme-review, etc.)
```
## Running tests
Tests are organized by fidelity level. Default `uv run pytest` runs everything
except `smoke` and `performance` (see `pyproject.toml` markers).
| Kind | Location | What it checks | Command |
| ----------- | -------------------- | ------------------------------------------------------- | -------------------------------------------------------------------- |
| Structural | `tests/structural/` | Files, dirs, frontmatter — cheap existence checks | `uv run pytest tests/structural/` |
| Behavioral | `tests/behavioral/` | Skill body content and protocol rules | `uv run pytest tests/behavioral/` |
| Integration | `tests/integration/` | Repo-wide references and cross-file consistency | `uv run pytest tests/integration/` |
| Harness | `tests/test_*.py` | `settings.json`, permissions, hooks, cost extraction | `uv run pytest tests/test_*.py` |
| Bats | `tests/*.bats` | Shell scripts (`checkpoint-archive`, `sync-theme`) | `bats tests/test_checkpoint_archive.bats tests/test_sync_theme.bats` |
| Smoke | `tests/smoke/` | Live skill invocation (costs tokens, non-deterministic) | `uv run pytest -m smoke` |
| Performance | `tests/performance/` | Stress matrix (18 combos, expensive, non-deterministic) | `uv run pytest -m performance` |
Run everything cheap and deterministic:
```shell
uv run pytest
```
Run the full suite including live/expensive tests:
```shell
uv run pytest -m ""
```
### Parallel execution
For the slower suites (`performance/`, `smoke/`, or `-m ""`), run tests in
parallel with [`pytest-xdist`](https://pytest-xdist.readthedocs.io/):
```shell
uv add --dev pytest-xdist # one-time install
uv run pytest -n auto # auto = one worker per CPU core
uv run pytest -n 4 -m performance # pin worker count for the 18-combo stress matrix
```
Skip `-n` for the default suite — it runs in ~0.5s, and xdist worker startup
adds more overhead than it saves.
## Privacy
`claude-damn` runs entirely on your local machine. The plugin itself does not
collect, transmit, store, or share any user data with the author or any
third-party service. Specifically:
- **Skills and hooks execute locally.** Every skill, every hook, every script
runs inside your Claude Code session on your own filesystem. Nothing is phoned
home to a `claude-damn` server — there is no `claude-damn` server.
- **Cost-tracking artifacts stay on-disk.** `/cost_` writes session token-usage
logs to `~/.claude/cost-log/` and `~/.claude/COST-SUMMARY.md`. `/cost-opt`
reads and compacts those same files. None of this is uploaded anywhere.
- **Checkpoint state stays in-repo.** `/checkpoint-save` writes `CHECKPOINT.md`
to your working tree and archives to `.checkpoints/` — local files only.
- **Tesseract transmissions are local-only.** `/tesseract` reads and writes
`~/.tesseract/shelf/*.md` and `~/.tesseract/bulk-beings.md` on your disk. The
skill explicitly advises gitignoring `~/.tesseract/` entirely in any repo that
mirrors `~/.claude`.
**Upstream data flows the plugin does not govern:**
- Your prompts and Claude's responses travel to Anthropic as part of normal
Claude Code operation. That is governed by
[Anthropic's privacy policy](https://www.anthropic.com/legal/privacy), not
this plugin.
- Skills that shell out to `gh`, `git`, `curl`, or other tools follow the
privacy practices of those tools and any remote services they talk to.
If any future version of `claude-damn` adds data collection or transmission,
this section will be updated before that version ships.
## Bonus Track: Jetbrains Configs
Prettier `Run for Files` setting:
```
**/*.{js,ts,jsx,tsx,cjs,cts,mjs,mts,json,vue,astro,md,mdx,toml,yaml,yml}
```
`Compare Files` tool canonical vs. active-dev:
`!.idea/*&!projects/*&!backups/*&!cost-log/*&!cache/*&!file-history/*&!ide/*&!paste-cache/*&!plugins/*&!serena-dashboard-static/*&!session-env/*&!sessions/*&!tasks/*&!shell-snapshots/*&!telemetry/*&!security*warnings*.json*&!history.jsonl*&!settings.json*&!!tests/fixtures/settings.json*&!policy-limits.json*&!.worktrees/*&!worktrees/*&!.checkpoints/*&!.remember*&!CHECKPOINT.md*&!.claude*&!docs/superpowers*&!**pycache**/*&!*.py[cod]*&!.venv/*&!\*.egg-info/!.idea/*&!projects/*&!backups/*&!cost-log/*&!cache/*&!file-history/*&!ide/*&!paste-cache/*&!plugins/*&!serena-dashboard-static/*&!session-env/*&!sessions/*&!tasks/*&!shell-snapshots/*&!telemetry/*&!security*warnings*.json*&!history.jsonl*&!settings.json*&!!tests/fixtures/settings.json*&!policy-limits.json*&!.worktrees/*&!worktrees/*&!.checkpoints/*&!.remember*&!CHECKPOINT.md*&!.claude*&!docs/superpowers*&!**pycache**/*&!*.py[cod]*&!.venv/*&!*.egg-info/*&!brainstorming/*&!usage-data/*!.idea/*&!projects/*&!backups/*&!cost-log/*&!cache/*&!file-history/*&!ide/*&!paste-cache/*&!plugins/*&!serena-dashboard-static/*&!session-env/*&!sessions/*&!tasks/*&!shell-snapshots/*&!telemetry/*&!security*warnings*.json*&!history.jsonl*&!settings.json*&!!tests/fixtures/settings.json*&!policy-limits.json*&!.worktrees/*&!worktrees/*&!.checkpoints/*&!.remember*&!CHECKPOINT.md*&!.claude*&!docs/superpowers*&!**pycache**/*&!*.py[cod]*&!.venv/*&!\*.egg-info/!.idea/*&!projects/*&!backups/*&!cost-log/*&!cache/*&!file-history/*&!ide/*&!paste-cache/*&!plugins/*&!serena-dashboard-static/*&!session-env/*&!sessions/*&!tasks/*&!shell-snapshots/*&!telemetry/*&!security*warnings*.json*&!history.jsonl*&!settings.json*&!!tests/fixtures/settings.json*&!policy-limits.json*&!.worktrees/*&!worktrees/*&!.checkpoints/*&!.remember*&!CHECKPOINT.md*&!.claude*&!docs/superpowers*&!**pycache**/*&!*.py[cod]*&!.venv/*&!*.egg-info/*&!brainstorming/*&!usage-data/*&!.credentials.json`
## License
[MIT](LICENSE)