---
title: QM
description: How ClawMetry observes QM — where its data lives, the environment variables that move it, what cost data is real, and the traps in its store.
eyebrow: Runtime reference
---

# QM

qm is a meta-orchestrator: one Slack and web surface per org that delegates to Pi, OpenCode, Codex or Claude Code interchangeably. The value ClawMetry adds above the child adapters is the scope layer — who in the org ran what, cron health across scopes, and per-run cost the children cannot see.

| | |
|---|---|
| **Runtime id** | `qm` |
| **Category** | Multiplayer agent harness |
| **Vendor** | Y Combinator |
| **Tier** | Paid tier (Starter and up) |
| **Upstream** | [qm.ycombinator.com](https://qm.ycombinator.com) |

## Where the data comes from

ClawMetry reads these locations, in order, and never writes to any of them.

- `$DATABASE_URL` — a PostgreSQL database, not a local file
- `Tables read: sessions, session_entries, session_llm_requests, tasks`

**Format.** PostgreSQL. Connections are short-lived and every one sets `default_transaction_read_only`.

## Environment variables

| Variable | Effect |
|---|---|
| `CLAWMETRY_QM_DATABASE_URL` | read-replica or alternate connection string |
| `DATABASE_URL` | qm's own env, used as the fallback |

See [Environment variables](/docs/config/environment/) for the variables that apply to every runtime.

## Tokens and cost

**Tokens.** Per-run, from `session_llm_requests`, normalised across the Anthropic and OpenAI usage shapes.

**Cost.** Derived from that per-run usage.

:::note Why this section exists
Runtimes differ enormously in what they persist. Rather than showing a plausible number everywhere, ClawMetry reports what the store actually contains and marks the rest unknown. A cost of `$0.00` and a cost of *unknown* mean different things, and the dashboard keeps them different.
:::

## Sub-agents and delegation

Each delegation is a durable `tasks` row plus a matching tool call and result in `session_entries`. Children carry zero tokens because qm keeps no per-child split — that is reported as unavailable rather than invented.

## Things worth knowing

- A Postgres driver is required. ClawMetry prefers pure-Python `pg8000`; with no driver installed, detection reports not-detected with a note rather than failing.
- `sessions.forked_from_session_id` is conversation fork provenance, not delegation. It never becomes a parent id.

## Verify it is being read

```bash
# Is this runtime detected, and would this install unlock it?
clawmetry runtimes --json | jq '.runtimes[] | select(.id=="qm")'

# Why is it locked, if it is?
clawmetry runtimes --why qm

# Any sessions ingested yet?
curl -s 'http://localhost:8900/api/local/sessions?limit=5' \
  | jq '.rows[] | {session_id, agent_id, events, cost_usd}'
```

If detection succeeds but no sessions appear, work through [Troubleshooting](/docs/ops/troubleshooting/) — the usual causes are a relocated data directory (set the override above), a session limit that is hiding older runs, or a runtime that simply has not written anything yet.