Diagnostics with doctor¶

veska doctor is the operator's window into the daemon. It's the one surface you reach for when something looks wrong - or to confirm things are healthy.

Start here¶

veska doctor status     # overall health rollup across all subsystems

Add --json for machine-readable output, or -v / --verbose to include failed queue rows inline.

Subsystem checks¶

Each subsystem has its own focused check:

Command	What it reports
`veska doctor status`	Overall health rollup across all subsystems
`veska doctor service`	OS-service health (installed? running?)
`veska doctor storage`	Filesystem storage metrics for `~/.veska/`
`veska doctor embedder`	Which embedder is elected, and its health
`veska doctor post_promotion_queue`	Backlog of post-commit embed/check work
`veska doctor pipelines`	Promotion pipeline health
`veska doctor config`	Effective configuration
`veska doctor identity`	Actor-attribution / identity tier
`veska doctor egress`	Outbound-connection posture
`veska doctor backup`	Backup status
`veska doctor wiki_render`	Wiki render health
`veska doctor savings`	Inline-snippet token savings (also `veska savings`)
`veska doctor bundle`	Bundle the above into one diagnostic report

When things look stuck¶

Search returns [] or degraded_reasons - embeddings are still catching up. Check veska doctor post_promotion_queue and eng_get_status's pending_embeds. See Semantic search & embeddings.
Daemon won't stay up - check veska doctor service, then the logs at ~/.veska/logs/daemon.log. If a crash-loop guard tripped:
```
veska doctor reset-crash-loop
```
Collecting a report for a bug - veska doctor bundle gathers the full diagnostic set in one shot.

doctor is the operator surface

By design, veska doctor is the thing an operator actually sees. If a subsystem can fail, it has a doctor check - start there before reading logs.