Jules Prompts

A curated library of machine-readable prompts, Agent Skills, and planted-failure fixtures to help coding agents turn intent into verifiable work. The procedures are harness-agnostic.

This site lists the Tasks. Skills are the same procedures in the format agents load in 2026. Fixtures are how those procedures go red.

Browse the Tasks to get started. Standing doctrine for a project’s AGENTS.md is in harness/AGENTS.md.

Tools these prompts pair with

A prompt tells an agent what to do. These do part of the same work mechanically, so the agent spends its judgement on the cases that need it.

• docproof reads a repository’s documentation and reports the claims its code contradicts: paths that no longer exist, commands that are not there, versions that moved. It is the automated half of Prove the docs, and it runs as a GitHub Action. MIT.
• rigout is an MCP server that lets an authorised agent use a real machine: commands, files, Docker, persistent terminals. pip install rigout.

Writing

Notes from actually running this kind of tooling now live at log.wecanuseai.com, so this stays a prompt library and the writing has somewhere of its own to accumulate. Most recently, why a clean verdict over five per cent of a repository is not a clean repository.