agent-sop
llms.txt74 starsChanged 3 days agoMITView on GitHub
agent-sop is a loadable SOP skill for coding agents (Claude Code, Codex CLI, Gemini CLI and others) that discovers the agent CLIs installed on your machine, assigns writer/reviewer/scout roles by capability for multi-agent work, and enforces cross-vendor review, evidence-based verification and human signoff (human-in-the-loop gates) before anything ships. The repository is a Markdown skill plus one standard-library Python script. Agents load SKILL.md; the workflow is agent-agnostic and never lets an agent publish or deploy on its own.
In this repositoryOne other file it gives its agents.
- sopSkill
The file
Files
# agent-sop > agent-sop is a loadable SOP skill for coding agents (Claude Code, Codex CLI, Gemini CLI and others) that discovers the agent CLIs installed on your machine, assigns writer/reviewer/scout roles by capability for multi-agent work, and enforces cross-vendor review, evidence-based verification and human signoff (human-in-the-loop gates) before anything ships. The repository is a Markdown skill plus one standard-library Python script. Agents load `SKILL.md`; the workflow is agent-agnostic and never lets an agent publish or deploy on its own. ## Docs - [SKILL.md](https://github.com/ythx-101/agent-sop/blob/main/SKILL.md): the workflow itself, written to be loaded by an agent as a skill. - [references/roster-protocol.md](https://github.com/ythx-101/agent-sop/blob/main/references/roster-protocol.md): roster discovery, condition-based slot assignment, reviewer-independence ladder, roster.json schema. - [references/review-checklist.md](https://github.com/ythx-101/agent-sop/blob/main/references/review-checklist.md): independent review checklist with the VERDICT/BLOCKING/SUGGESTED/EVIDENCE contract. - [templates/signoff-packet.md](https://github.com/ythx-101/agent-sop/blob/main/templates/signoff-packet.md): Markdown signoff packet template. - [scripts/discover-roster.py](https://github.com/ythx-101/agent-sop/blob/main/scripts/discover-roster.py): agent CLI discovery with optional live health probes (Python 3.8+, standard library only). - [README.zh.md](https://github.com/ythx-101/agent-sop/blob/main/README.zh.md): Chinese README. ## Key concepts - Capability slots: orchestrator, sole-writer, reviewer, scout, monitor. Assigned by condition, not by agent name; exactly one writer per repo at a time. - Workflow tiers: high-risk work follows the full sequence (research, plan, human approval, implementation, independent review, verification, signoff packet, human signoff, deploy); low-risk work uses a lighter path (SKILL.md section 1.1). - Gates: publish, deploy and data-write steps are never self-served by an agent; approval must be explicit and signoff can only be set to approved by a human. - Review contract: independent reviewers answer with VERDICT, BLOCKING, SUGGESTED and EVIDENCE.
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.