mirror of
https://github.com/boshu2/agentops.git
synced 2026-09-14 15:08:13 +08:00
Bo
a6359795bf
Keep fresh author-distinct validation mandatory while making verdict and report persistence consumer-driven. Align the RPI/Validate contracts, executable behavior, current guidance, regression coverage, and generated projections; preserve the Gas City 1.4 cutover.
1.2 KiB
1.2 KiB
Newcomer guide
AgentOps is a skill-based evidence loop for agent-created work. Start with one small behavior and use RPI when you want the full loop.
What to expect
- Plan makes acceptance and scope explicit.
- Implement performs one bounded experiment.
- A fresh validator judges exact content.
- The result records checked and unchecked claims; durable storage is optional.
- The invocation stops.
The one thing you do need is a coding agent to run the skills in: AgentOps
skills execute inside your agent runtime (Claude Code, Codex, Cursor, …), and
/rpi is typed in that agent's chat. You do not need a tracker, Git
repository, multi-agent substrate, or ao binary for the semantic loop. If
your project has those systems, keep using their native workflow after
AgentOps reports its verdict.
Use premortem when a plan deserves an advisory challenge, council when multiple independent views justify the cost, and postmortem only for an explicit causal question.