oh-my-ai/agent-init
Creates, updates, or prunes an AGENTS.md for any repository by auditing the codebase, detecting non-discoverable gaps, and drafting minimal high-signal instructions that agents cannot infer from reading the code.
90
94%
Does it follow best practices?
Impact
78%
1.06xAverage score across 3 eval scenarios
Passed
No known issues
Evaluation results
80%
8%Set Up Agent Instructions for Orbital API
Discoverability filter and audit trail
Exact heading structure
0%
75%
Scope statement present
0%
0%
CI flag documented
100%
100%
Undocumented env var documented
100%
100%
Deprecated module status documented
100%
100%
No tech stack content
80%
100%
No directory overview
100%
100%
No linter duplication
100%
100%
Package manager mismatch documented
100%
100%
Audit trail file created
100%
100%
Correct section headings used
0%
0%
85%
Refresh Agent Instructions for Proxima Backend
Incremental AGENTS.md update with pruning
Changes summary file created
100%
100%
Removals documented with reasons
100%
100%
Tech stack section removed
100%
100%
Stale command removed
100%
100%
Generic advice removed
100%
100%
Non-discoverable content preserved
100%
100%
New finding added
0%
0%
Correct heading structure maintained
50%
50%
Redundant directory overview removed
100%
100%
Tooling-enforced rules not duplicated
100%
100%
70%
8%Set Up Agent Instructions for FinStack Monorepo
Monorepo hierarchical AGENTS.md and targeted questions
Hierarchical structure recommended
50%
41%
No monolithic file for all modules
100%
100%
No generic discoverable questions
80%
100%
Targeted ambiguity question asked
41%
66%
Question count within limit
0%
0%
CI vs local difference documented
100%
100%
Do-not-touch constraint documented
100%
100%
No tech stack summaries in output
37%
62%
Scope statement present
0%
0%
Audit trail created
83%
100%
- Evaluated
- Agent
- Claude Code
- Model
- Claude Sonnet 4.6