Menu
Independently verify that an HMG integration can write, recall, handoff, preserve scope isolation, and emit lifecycle events.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Produce an HMG health scorecard from doctor JSON, disposable memory smoke, agent-event explain, and daemon latency sampling.
Create a non-invasive HMG integration plan for a target repository using MCP, hooks, runtime, and smoke tests behind a feature flag.
Use HMG durable coding-agent memory when a task benefits from branch-aware recall, prior decisions, correction history, governance, or an end-of-task handoff.
Manually load scoped HMG context when SessionStart or prompt-time hooks are unavailable.
Run the local mem0 OSS vs HMG branch-isolation scorecard and interpret latency, hit rate, scope leakage, and agent brief usefulness.
Plan safe dry-run imports from mem0, MEMORY.md, Claude memory banks, or project documents into HMG.
| name | hmg-test-integration |
| description | Independently verify that an HMG integration can write, recall, handoff, preserve scope isolation, and emit lifecycle events. |
Use this workflow after an integration plan or init has been applied.
hmg agent-event explain.Use the deterministic runner for the handoff artifact:
python3 scripts/hmg-integration-scorecard.py test --agent <id> --target <repo> \
--baseline-command '<existing flag-off test command>'
It writes .hmg-integration/scorecard.json with command exit state and hashed
evidence, never raw command output. Host loaded, executed, emitted, and
acknowledged stages must still be interpreted separately from
hmg integrations explain.
The result should prove memory write, recall, handoff, and scope isolation, not merely that config files exist.