mirror of
https://github.com/hanzoai/kai.git
synced 2026-10-17 11:18:06 +00:00
zooqueenandHanzo Dev
b00044ece2
Every runner printed whatever came back and exited 0, and fell back to a second raw HTTP call when the SDK import failed. Now each recipe carries expect.json, the answers a careful reader gives on its own input: a list of acceptable options for a choice, of level indices for a score, true or false for a noul. run.py and run.ts send request.json through the SDK, print every answer, and exit 1 when a checked one is wrong. A question with no obvious answer on its input is printed and not checked. Recipes 09 and 11 had no input with an obvious answer, so each state gains the one fact that decides it. The recipe READMEs describe what each recipe asks instead of capabilities it never exercised, and copy the request and checked answers from the two JSON files. Co-authored-by: Hanzo Dev <[email protected]>
Agent Progress & Loop Detection
Ask Kai whether an agent that keeps hitting the same failure is stuck, and what the runtime should do next.
Request (request.json)
{
"model": "kai",
"state": {
"goal": "Make all vitest tests pass in src/lib/auth",
"history": [
"turn 1: ran tests -> 2 failed",
"turn 2: edited auth.ts -> ran tests -> 2 failed (same error)",
"turn 3: edited auth.ts -> ran tests -> 2 failed (same error)",
"turn 4: edited auth.ts -> ran tests -> 2 failed (same error)"
]
},
"questions": {
"is_stuck": {
"type": "noul",
"instructions": "Is the agent caught in an unproductive loop?"
},
"next_strategy": {
"type": "choice",
"instructions": "What recovery action should the agent runtime take?",
"criteria": {
"continue": "allow another turn with existing strategy",
"rollback": "revert last changes to git HEAD and try fresh approach",
"escalate": "pause execution and ask user for guidance"
}
}
}
}
Checked answers (expect.json)
is_stuck(noul): P(true) above 0.5next_strategy(choice):rollbackorescalate
Run
From the repository root, after pip install -r requirements.txt and npm install:
curl -sS --fail-with-body https://api.hanzo.ai/v1/decisions \
-H "Authorization: Bearer $HANZO_API_KEY" \
-H "Content-Type: application/json" \
-d @recipes/04_agent_progress_eval/request.json
python3 recipes/04_agent_progress_eval/run.py
npx tsx recipes/04_agent_progress_eval/run.ts
run.py and run.ts print every answer and exit 1 when a checked answer is wrong.