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

# NanoClaw

NanoClaw gives each session its own pair of SQLite files — one the host writes and one the container writes — and reconstructs a transcript by merging them. Token data arrives indirectly, through the Claude Code state the host bind-mounts into the container.

| | |
|---|---|
| **Runtime id** | `nanoclaw` |
| **Category** | Containerised agent runtime |
| **Vendor** | Nanoco |
| **Tier** | Paid tier (Starter and up) |
| **Upstream** | [github.com/nanocoai/nanoclaw](https://github.com/nanocoai/nanoclaw) |

## Where the data comes from

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

- `<checkout>/data/v2-sessions/<group>/<session>/inbound.db` — host writes
- `<checkout>/data/v2-sessions/<group>/<session>/outbound.db` — container writes
- `<data>/v2-sessions/<group>/.claude-shared/projects/…` — mounted Claude state

**Format.** Two SQLite files per session; `seq` is unique across both tables (host writes even numbers, container odd), so a transcript is a merge sort on `seq`.

## Environment variables

| Variable | Effect |
|---|---|
| `CLAWMETRY_NANOCLAW_DIR` | point at a v2-sessions dir or a checkout root |

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

## Tokens and cost

**Tokens.** Not in the message tables. Real counts come from the bind-mounted Claude Code transcripts when the install is new enough to have them.

**Cost.** Derived from the mounted Claude transcripts. Installs older than the bind-mount degrade to a message-only view with no cost.

:::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

Through the mounted Claude state.

## Things worth knowing

- There is no `NANOCLAW_HOME`, no `DATA_DIR` and no `~/.nanoclaw`. The data dir is resolved relative to the install CWD, so the real path is `<checkout>/data/v2-sessions`. ClawMetry discovers it from a bounded set of common checkout locations.
- The runtime actively owns these files, so ClawMetry always opens them read-only and immutable.

## Verify it is being read

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

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

# 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.