Self-Diagnose

Read Sutando's own observable state (logs, git, memory, build log, pending questions, health check, cold-review log) over a chosen window and produce a structured narrative:

• What's been happening — significant events, PRs shipped, tasks processed, user interactions
• What's broken — errors in logs, pending questions, health-check failures, 1006/1011/1007 transport events, unresolved bugs surfaced in logs
• What I'd do next — concrete prioritized actions grounded in what was observed

Usage: /self-diagnose [--since 24h]

ARGUMENTS: $ARGUMENTS

Flow

1. Gather. Run bash skills/self-diagnose/scripts/gather.sh [window] — collects log tails, git log, build_log.md tail, pending-questions, health-check output, cold-review-log, and recent Discord activity into /tmp/sutando-diagnose-<ts>/.
2. Synthesize. Read each gathered file. Group findings under the three headings above. Be concrete: cite log lines, PR numbers, timestamps. Skip speculation.
3. Save. Write the report to notes/diagnose-YYYY-MM-DD-HHMM.md with frontmatter (title, date, tags: [diagnose, self]).
4. Summarize. Reply to the caller with a 5–10 line executive summary + path to the full report.

Cross-node mode (resolves #421)

When Sutando works on one machine but is broken on another, the helper script runs gather.sh on both sides over SSH and produces a structured comparison:

bash skills/self-diagnose/scripts/gather-remote.sh <ssh-target> [window]
# e.g.
bash skills/self-diagnose/scripts/gather-remote.sh mac-mini
bash skills/self-diagnose/scripts/gather-remote.sh user@macbook.local 6h

Output: /tmp/sutando-diagnose-cross-<ts>/{local,remote,diff.md} plus a persisted copy at notes/diagnose-cross-node-<YYYY-MM-DD>.md. The comparison report surfaces commit drift, health-check differences, voice-agent error counts, quota state, and PR-view divergence per side.

Override the remote sutando path via SUTANDO_REMOTE_REPO=/path/on/peer if the failing node's checkout isn't at the default ~/Desktop/sutando.

Security posture (per #421): read-only on the remote (only gather.sh runs there, no mutation), allowlist enforced by gather.sh's existing scope (no .env, no tokens), per-session SSH (no daemon, no persistent state on the remote).

Default window

24 hours if unspecified. Accept: 24h, 3d, 1w. Longer windows = broader scope, higher token cost.

What a good report looks like

• Cites specific log timestamps (18:28:16 — transport 1006 after NoteView injection) rather than "some errors happened"
• Lists PRs by number + status (#394 mergeable no reviews, #354 retention sweep open)
• Separates recurring from one-off issues
• Prioritizes "what I'd do next" by impact and blocking, not by what's most recent

What to skip

• Verbose log dumps — the gathered files are on disk already; just reference them
• Speculation without log evidence — "maybe the Gemini server is..." should be backed by a log line showing the close code and preceding event
• Restating CLAUDE.md / build_log.md content — those are already durable; only surface what's NEW in the window

Related

• notes/cold-review-capability.md — sister capability for PR-level review
• notes/voice-transport-1006-hypothesis.md — example of the depth of analysis a self-diagnose report should match
• build_log.md — the canonical "what has been built" log; complements but doesn't replace