| name | glean-status |
| description | Show glean pipeline health - pending queue, delivered fragments, outbox, recent activity, last extraction timestamp, and config summary |
| allowed-tools | Bash, Read |
| user-invocable | true |
glean-status
Show the current health of the glean extraction pipeline.
Steps
- Run the worker status command. The worker outputs a JSON object on stdout:
PYTHONPATH="${CLAUDE_PLUGIN_ROOT}/scripts" python3 -m glean.worker --status
-
Parse the JSON. Expect fields like: pending, processing, failed, delivered_7d, outbox, last_extraction, config (with output_mode, vault_path, vault_folder, extraction_model, prompt_version, max_fragments_per_session). If the actual JSON shape differs, adapt to what is returned and surface any unknown fields verbatim.
-
Render a compact human-readable status table. Suggested layout:
glean pipeline status
---------------------
Queue pending: N processing: N failed: N
Delivered last 7d: N total: N
Outbox N (items needing attention)
Last run 2026-04-05T18:45:00Z
Config
mode: obsidian
vault: /Users/ved/.../VedOS/Fragments
model: haiku
prompt_version: 1
max fragments: 3 / session
-
If outbox count > 0, list each outbox item with its fragment_id and reason (vault write failure, etc). Read files from ${CLAUDE_PLUGIN_DATA}/outbox/ only if the status JSON does not already include the details.
-
If failed count > 0, mention that .failed files exist in the queue dir and suggest the user inspect ${CLAUDE_PLUGIN_DATA}/worker.log.
-
Keep output tight - one screen, no decoration beyond what is shown above. Do not add emojis.
Error handling
- If the worker command fails (non-zero exit), show stderr and the exit code. Do not attempt to fabricate status.
- If
${CLAUDE_PLUGIN_DATA} is unset or the config is missing, report that clearly and suggest the user re-run the plugin install.
