Overview
The AI technical-debt scanner — a free, diagnostic-only CLI that scores your repo's agent harness. No writes, no account, no config required.
Docs
Everything about the harness
The scanner, the setup interview, the skills catalog, the config reference, and the
reasoning behind all of it. Start with a read-only scan:
npx @paniolo/cli scan .
paniolo scan
Guide
Quick start, install matrix, report reference, scoring tiers, and a CI cookbook.
Research
Every scanner feature traced to the paper and statistic behind it.
Calibration
Published threshold derivations — weak, normal, and mature anchors for each score.
Compare
A factual comparison with AgentLint, agnix, and AgentLinter — including what each does better.
Rule catalog
Every built-in rule with its default severity and scoring dimension — generated from the registry.
Setup
paniolo init
Create, adopt, or connect a harness around the repos you already have — a guided interview, a dry run before anything changes, and every answer repeatable as a flag.
Config reference
The paniolo.config.json every command reads — leaf and harness forms, repoDefaults, repoTypes, and precedence.
Knowledge base
LLM Wiki
The pattern, the research, and every wiki feature in the CLI — init, lint, page ops, config wiring, and qmd retrieval.
Agent surface
Skills
The skills catalog — audit, remediation, and setup skills plus the engineering library. One install covers every skill-capable agent.
Why
Methodology
The knowledge base as source code, the AHE evolution loop, core principles, and the three-layer model.