Setup Path
Can a new user find exact install, config, run, and smoke-test steps without guessing?
methodology / public evidence only
How MAXIM scores MCP and AI-agent readiness
MAXIM looks for public signals that a repo may be hard for agents or maintainers to install, configure, debug, or operate safely. The index is a triage aid, not a certification or endorsement.
Runtime Faults
Do public issues show failures around auth, transports, tool calls, subprocesses, or model sessions?
Schema And Tool Safety
Are tool inputs, expected outputs, permission boundaries, and failure modes documented clearly?
Agent Instructions
Does the repo expose AGENTS.md, CLAUDE.md, or equivalent operating notes for agent workflows?
Score meaning
- A: strong public signal that the issue is specific, current, and useful for readiness triage.
- B: useful public signal, but the repo may need more context before a paid diagnostic makes sense.
- Rows are filtered out when the evidence is stale, crowded, security-sensitive, or mostly internal workflow bookkeeping.
What maintainers can use now
The starter pack gives a copyable AGENTS.md template, readiness checklist, and sample report format. It is designed to be useful even if no paid work happens.