agent-ops-patterns
Maps of who decides what inside open-source agent systems, and the skill to make one of your own.
The gallery
Each map lists the recurring judgments in one system and who holds each (code, a model, or a person), with how it has failed and what checks it where the mapping found them, the prompt pieces and assembly paths identified in the mapped source for the seats a model holds, and a report of what the mapping exposed. The OpenHands map also quotes prompt source text on 14 of its cards, each quote linked to the lines it came from.
Its goal judge reads "complete": "false" as complete, because in Python bool("false") is True. We reproduced it with a failing test and reported it as #5492, which the project has labelled ready-for-dev. Two outside contributors have since opened fix PRs, #5493 and #5563 (one a draft); neither is merged yet (checked 2026-10-12). All the gaps →
All maps, how to read one, and what a map is not →
Make your own
The cog-ops-map skill has your agent map a system you work on. In Claude Code:
claude plugin marketplace add u00dxk2/agent-ops-patterns
claude plugin install agent-ops-skills@agent-ops-patternsThe method is "Cognitive Operations Maps".
Also in the repository
- A five-question audit to paste to your own agent: read-only, scored out of five.
- Small zero-dependency libraries and written patterns for running LLM agents in production.