Paniolo config init

paniolo-config-init — scaffold paniolo.config.json

Generate a starter paniolo.config.json. The scanner only reads this file — it never writes it — so the file the user ends up with is whatever this skill writes.

When to use

Use this skill when the user wants to create or set up a paniolo.config.json, or asks which AI vendors/harnesses a repo should support — for example: "init paniolo config", "set up paniolo.config.json", "which AI vendors should this repo support", or the /paniolo-config-init trigger.

Flow

1. Check for an existing config

Look for paniolo.config.json at the repo root. If present, read it and show the current harnesses value before continuing — don't silently overwrite a file someone already tuned.

2. Detect likely vendors from existing adapters

Glob for adapter markers already in the repo as a hint, not a decision:

Marker	Harness
CLAUDE.md, .claude/	claude
.cursor/, .cursorrules	cursor
.github/copilot-instructions.md	copilot
AGENTS.md, .codex/	codex
GEMINI.md, .gemini/	gemini
.agents/	antigravity

3. Ask which AI vendors/harnesses to support

Ask the user to pick from the six harnesses the scanner understands — claude, cursor, copilot, codex, gemini, antigravity — noting which ones Step 2 detected as likely defaults. If the agent supports a multi-select prompt, use it; otherwise ask in plain text and accept a comma-separated list.

If the user picks none, or says "all" / "no preference", leave harnesses out of the config entirely — an empty/absent list means "no declared scope," and every harness-specific rule still runs, the same as omitting --harness on the CLI.

4. Write the config

Write paniolo.config.json at the repo root, kept minimal — this config grows with the rule set, so don't pre-populate rules with every rule ID:

{
  "$schema": "https://paniolo.ai/schemas/paniolo.config.v1.json",
  "extends": ["@paniolo/cli/presets/meta-harness.json"],
  "harnesses": ["claude", "cursor"],
  "rules": {},
  "options": {}
}

• harnesses — exactly what the user picked, lowercase, matching the CLI's --harness flag and a report's harnesses field. Omit the key when the user chose "all" rather than writing an empty array.
• extends — default to @paniolo/cli/presets/meta-harness.json unless the user names a different preset (paniolo-reference, scan-legacy, guidance-strict, boundary-minimal).

5. Confirm and offer a scan

Show the written file's contents. Offer to run npx --yes @paniolo/cli scan --format json . (the paniolo-scan skill) to confirm the new config loads and to get a baseline score.

Guardrails

• Never invent a harnesses value the user didn't pick or explicitly confirm from a detected default.
• Don't write rules, guidance, or boundary overrides unless asked — this skill scaffolds, it doesn't tune.
• Never overwrite an existing paniolo.config.json without showing its current contents and getting explicit confirmation first.

---

About Paniolo

Paniolo builds precision infrastructure for autonomous engineering — the harness layer around your coding agents: project intelligence, observability, guardrails, and the structural patterns that turn generated code into production-grade output.

paniolo.config.json is how a repo tells paniolo scan (via @paniolo/cli) which harnesses it supports and how to weigh its rules. This skill scaffolds that file; the scanner reads it.