Mindburn LabsLab behind HELM
Try OSS

INTEGRATE_IN_5_MIN

Integrate HELM in 5 Minutes

No trust required. Every step produces verifiable output.

1. Start HELM

docker compose up -d
curl -s http://localhost:8080/healthz   # → OK

2. Swap base_url (Python)

- client = openai.OpenAI()
+ client = openai.OpenAI(base_url="http://localhost:8080/v1")

3. Make a tool call

import openai

client = openai.OpenAI(base_url="http://localhost:8080/v1")
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "List files in /tmp"}]
)
print(response.choices[0].message.content)

4. See a deny

curl -s http://localhost:8080/v1/tools/execute \
  -H 'Content-Type: application/json' \
  -d '{"tool":"unknown_tool","args":{"bad_field":true}}' | jq .

Expected:

{
  "status": "DENIED",
  "reason_code": "ERR_TOOL_NOT_FOUND",
  "receipt_hash": "sha256:..."
}

5. Export EvidencePack

./bin/helm export --evidence ./data/evidence --out pack.tar
ls -la pack.tar

Expected: deterministic pack.tar file.

6. Verify offline

./bin/helm verify --bundle pack.tar

Expected: verification: PASS — no network required, air-gapped safe.

Done. You now have a governed tool-call loop with cryptographic receipts and offline-verifiable evidence.

Next: