Harden Skill Endpoint
Run the autonomous skill-hardening loop once: build/grow the harness, then each round propose -> prove on the held-out slice -> keep ONLY if strictly better (compounding), and stage ONE pending_version for the user's Apply. By-id-direct; cross-space. See docs/design/SKILL_AUTONOMOUS_HARDENING_LOOP.md.
Path Parameters
skill_id*Skill Id
Query Parameters
max_rounds?Max Rounds
Response Body
application/json
application/json
curl -X POST "https://example.com/skills/string/harden"{}{
"detail": [
{
"loc": [
"string"
],
"msg": "string",
"type": "string",
"input": null,
"ctx": {}
}
]
}