Deeds — AI skill for Claude Code
A CLI and Claude Code plugin that reads every commit's diff and counts what actually got better: new capabilities, fixes, and upkeep.
How to install Deeds
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open danielmiessler/Deeds and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Deeds does
A CLI and Claude Code plugin that reads every commit's diff and counts what actually got better: new capabilities, fixes, and upkeep. A way to measure work done with AI that doesn't depend on counting PRs.
Alternatives in AI
- Impeccable — The design language that makes your AI harness better at design 13.9k ★
- Better Harness — Review this project's AI coding workflow with Better Harness and generate a durable report 2.1k ★
- VibeMeter — A project to measure costs for Cursor and other AI provider 379 ★
README
Deeds
Measure the work in a code repo by what got better, not by how many PRs were opened.
Pull requests and commit counts measure activity. They show that code changed, not whether the product got better. A refactor, a revert and a typo fix count the same as a new feature, and someone building alone with AI agents may never open a PR at all.
Deeds reads the diff of every commit in a window and asks what that change did to the product. A change that made it better is a **deed**, and every deed is one of three kinds:
| Kind | What it means | Example |
|---|---|---|
| cap | A capability gained or deepened. The product can do something it couldn't before, or does an existing thing more fully. | added CSV export to reports |
| fix | Something moved from broken to working. Security fixes count here. | stopped logging session tokens |
| tend | Upkeep users never see: refactors, dependency bumps, performance work, tests. | bumped hono to 4.6 |
Renames, reverts and busywork produce no deeds. Commit messages are never read, so a commit called "feat: X" only counts as a cap if its diff adds one. The three counts stay separate and are never blended into one score. The argument for all of this is at [workdeeds.ai](https://workdeeds.ai).
What you get
A real run over the last 90 days of [golang/go](https
Related Skills
RayCodes AnatomySkill
Anatomy Skill 1.1 Just Got Better: Turns Agents Into 3D AI Illustrator! Turn code and technical concepts into
MCP Notebooklm Go
Unofficial Go API and agentic skill for Google NotebookLM. Full programmatic access to NotebookLM's features—i
My Jupyter Lab
Python API and agentic skill for Google Gemini Notebook. Full programmatic access to NotebookLM's features—inc
Shuvbooklm
Unofficial Python API and agentic skill for Google Gemini Notebook. Full programmatic access to NotebookLM's f
Better Than Claude Skills
Expand your LLM and AI agent capabilities with this collection of modular tools and specialized skills.
Cc Judge
Shared LLM-evals framework for running Claude Code agents in containers, scoring outcomes with an LLM judge, a
Related Agents
Execution
Reviews whether the task actually got done — plan followed, work finished rather than finished-looking, nothin
Claim Grader
Use when a contributor wants a soft-gate review of a commit message, PR body, sprint section, or journal entry
Technical Verifier
Adjudicates an analyzer's claims against implementation code, manifests, and tests — counting files rather tha
