footprint
Skill18 starsChanged 4 days agoMITView on GitHub
Local SLM that learns from Claude Code transcripts ("footprint") and replays the project's working style. Trigger on /footprint, "take footprint", "train footprint", "footprint status", "footprint run <task>".
1 other project uses a version of this.
Adopted by
Most recently taken up by
- somasundarv/footprinttheir own version, first seen Jul 4, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- somasundarv/footprint★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository2 other files it gives its agents.
- UI-Web/AGENTS.mdAGENTS.md
- UI-Web/CLAUDE.mdCLAUDE.md
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 35
Always run commands from /Users/aman/Developer/footprint.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Footprint
Footprint collects this machine's Claude Code session transcripts, fine-tunes a
small local model on them (MLX LoRA on Apple Silicon, torch+PEFT elsewhere),
and serves it as an OpenAI-compatible API for OpenCode/Cursor/Codex.
Code layout: footprint.py (entry shim) -> footprint_core/ (cli, config,
transcripts, service, backends/{mlx,torch}_backend.py).
Repo: /Users/aman/Developer/footprint. Python: /Users/aman/Developer/footprint/.venv/bin/python
(if .venv missing, run setup first with system python3).
Commands
Map the user's /footprint <arg> to one Bash call:
| Invocation | Run |
/footprint or /footprint status | .venv/bin/python footprint.py status |
/footprint setup | python3 footprint.py setup |
/footprint trace | .venv/bin/python footprint.py trace (arm BEFORE chatting; only sessions after this become training data) |
/footprint collect | .venv/bin/python footprint.py collect <current project dir> |
/footprint collect all | .venv/bin/python footprint.py collect --all-projects-nonexistent-dir (falls back to all projects) |
/footprint train | .venv/bin/python footprint.py train (long; run in background) |
/footprint install | .venv/bin/python footprint.py install (Ollama model footprint + OpenCode /footprint command — user never runs serve) |
/footprint context | python3 footprint.py context (preview previous-session state for this project) |
/footprint context install | python3 footprint.py context install (SessionStart hook in ~/.claude/settings.json: new sessions continue from the last one; OpenCode plugin auto-injects + /continue command; reads both Claude and OpenCode sessions) |
/footprint switch | python3 footprint.py switch (import latest Claude chat for this project into OpenCode with full history; prints opencode -s <id>) |
/footprint serve | .venv/bin/python footprint.py serve (rebuild Ollama model footprint; install / adapter use already do it) |
Always run commands from /Users/aman/Developer/footprint.
Flow for a fresh user
setup— venv + mlx-lm + skill symlink.trace— arm tracing FIRST, before chatting with Claude. Writes~/.claude/footprint-trace; collect then only uses sessions newer than the marker (delete marker to use all history).- Chat with Claude Code normally — sessions land in
~/.claude/projectsautomatically. collect— parses~/.claude/projects/<project>/*.jsonlintodata/train.jsonl+data/valid.jsonl(chat format, tool calls encoded as<tool name="...">{args}</tool>).train— LoRA onmlx-community/Qwen2.5-Coder-1.5B-Instruct-4bit(override withFOOTPRINT_MODEL, iterations withFOOTPRINT_ITERS, default 300). Downloads ~1 GB first time. Run in background, report loss when done.install— the user never runs the server. The active adapter is converted to GGUF and built into Ollama modelfootprint(baseqwen2.5-coder:1.5b); Ollama serves the OpenAI-compatible API athttp://127.0.0.1:11434/v1and starts at login. Needs Ollama installed and running. Also installs into OpenCode: providerfootprint+/footprint <task>command (~/.config/opencode/commands/footprint.md) so typing/footprintthere makes OpenCode work like Claude. Other tools point at the always-on server:- Cursor: Settings → Models → OpenAI base URL
http://127.0.0.1:11434/v1, any API key, add modelfootprint. - Codex CLI: set a custom provider with that base URL in
~/.codex/config.toml.
- Cursor: Settings → Models → OpenAI base URL
Report command output tersely. If mlx-lm import fails, rerun setup.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

