What this gives you

Discover the installed CLI surface and record version-scoped command evidence
Inspect configuration, gateway health, channels, and logs without exposing credentials
Classify a fault before making one least-change, reversible repair
Produce a concise operator handoff with evidence, outcome, and rollback condition

Preflight

Before you start

A working OpenClaw installation and basic shell literacy

Syllabus

Work through the map

Discover commands without copying a command dump20 min
Inspect configuration without exposing secrets20 min
Separate doctor guidance from gateway health20 min
Diagnose channels and logs without creating a second incident20 min
Plan a safe update instead of using update as a repair button20 min
Recover the diagnostic path and hand off cleanly20 min
Capstone: ten-minute operator triage20 min
Final assessmentAfter coursework

Course details and supporting evidence

What you will do

Operate a disposable or approved non-production instance as an evidence collector before you become a repairer. Each lab ends with a small, redacted record: command, timestamp, question answered, result, and next safe move. A green process is not permission to connect a real recipient or make an external change.

This course preserves the legacy /courses/openclaw-cli-essentials/ route. It does not replace or redirect that route.

Course map

Discover the installed command surface and pin the version that supplied it.
Read configuration without turning secrets into evidence.
Separate local setup, gateway health, and channel runtime signals.
Use scoped channel probes and bounded log captures to classify a fault.
Plan an update with a before/after gate instead of treating update as recovery.
Produce a ten-minute triage report and escalate when one repair is not enough.

Original diagnostic map

symptom
  │
  ├─ CLI unavailable? ──> version + help evidence ──> install/PATH owner
  ├─ config invalid? ───> config validate / doctor ──> smallest config repair
  ├─ gateway absent? ───> gateway status / health ──> service owner
  ├─ channel degraded? ─> channels status --probe ──> account-specific evidence
  └─ delivery unclear? ─> bounded logs + dead letters ─> preserve and escalate

Before every write: record state → preview/dry-run → one change → validate → rollback condition.

Optional orientation video — pending editorial approval

freeCodeCamp.org, OpenClaw Full Tutorial for Beginners – How to Set Up and Use OpenClaw (ClawdBot / MoltBot) (n1sfrc-RjyM) is an optional orientation, not command authority. The candidate was public and embed-eligible when curated, with automatic English captions reported. An editor must watch and cross-check 02:44–03:49 (gateway concept), 06:58–09:40 (configuration context), 16:11–17:46 (audits and health checks), and 50:52–54:26 (sandbox validation) against the linked primary receipts before embedding. Learners can complete every exercise without video.

Completion evidence

Submit a redacted diagnostic packet: installed version, one help excerpt, config validation result, health or status result, a targeted channel/log observation if applicable, the one selected repair or reason for no change, post-check, rollback condition, and handoff owner. Do not include tokens, raw secret values, private endpoints, personal message content, or full unbounded logs.

Checkpoints

A command returns a cached health snapshot. What is the next action before declaring a channel healthy? Run a scoped live probe and state its account boundary.
config validate fails. What must precede a config write? Identify the stated error and use a preview or dry-run for the smallest correction.
A dead letter exists. What must happen before resubmission? Correct the underlying fault and preserve the event ID plus failure evidence.

Review course maintenance history →