AI for Developers
Agent skill (SKILL.md) Open source MIT 1.1k

Chisle

Cut your AI coding agent's token bill on three axes: terse prose, YAGNI-first code, and tool-output compression. Claude Code, Pi, Cursor, Codex, Gemini + 4 more. Zero deps, published benchmarks inc...

Install

For this project
$ npx skills add JayPokale/Chisle -a claude-code
For every project
$ npx skills add JayPokale/Chisle -a claude-code -g

Or copy the skill folder into .claude/skills/ (project) or ~/.claude/skills/ (all projects).

For this project
$ npx skills add JayPokale/Chisle -a codex
For every project
$ npx skills add JayPokale/Chisle -a codex -g

Or copy the skill folder into .agents/skills/ (project) or ~/.agents/skills/ (all projects).

For this project
$ npx skills add JayPokale/Chisle -a cursor
For every project
$ npx skills add JayPokale/Chisle -a cursor -g

Or copy the skill folder into .agents/skills/ (project) or ~/.cursor/skills/ (all projects).

For this project
$ npx skills add JayPokale/Chisle -a gemini-cli
For every project
$ npx skills add JayPokale/Chisle -a gemini-cli -g

Or copy the skill folder into .agents/skills/ (project) or ~/.gemini/skills/ (all projects).

For this project
$ npx skills add JayPokale/Chisle -a github-copilot
For every project
$ npx skills add JayPokale/Chisle -a github-copilot -g

Or copy the skill folder into .agents/skills/ (project) or ~/.copilot/skills/ (all projects).

See the repository README for installation instructions.

When to use

Use it when agent replies and generated code run long and you want shorter answers and smaller tool output.

Chisle tries to cut an agent's token bill along three axes: terse prose, YAGNI-first code and compression of tool output. Install with npx chisle. It has no dependencies and works in Claude Code, Pi, Cursor, Codex, Gemini and several more agents.

The README's example prompt asks for an order state machine with enforced transitions. With no ruleset the answer is 121 lines and 1,186 answer tokens, and with Chisle it's 62 lines and 530 tokens, with the verbatim answers committed in the repo's benchmarks. It claims fewer tokens than Caveman and Ponytail with no accuracy lost, which is the author's benchmark.

Related skills

All skills