UNPKG

openclaw

Version:

Multi-channel AI gateway with extensible messaging integrations

172 lines (134 loc) • 8.15 kB
--- summary: "Run doctor, pick a mode, and read the read-only lint report" title: "Run doctor" read_when: - You want to run doctor and choose the right flags - You need a read-only health report for CI or preflight automation --- Run `openclaw doctor` to repair and migrate an OpenClaw install. This page covers the command, its automation flags, and the read-only lint mode. When the System agent runs Doctor, it uses a separate process from the same OpenClaw installation. Bulk diagnostic checks can then run without blocking the Gateway's event loop. This uses the existing `--non-interactive` behavior, including its safe migrations; it does not enable additional repairs. Once started, Doctor finishes and releases its resources before a cancelled caller settles, so cancellation cannot abandon an in-progress migration. ## Quick start ```bash openclaw doctor ``` ### Headless and automation modes <Tabs> <Tab title="--yes"> ```bash openclaw doctor --yes ``` Accept default non-service repairs without prompting and enter maintenance under the [service-preservation and installation-drift rules](/cli/doctor/recovery#gateway-service-recovery). </Tab> <Tab title="--fix"> ```bash openclaw doctor --fix ``` Apply recommended non-service repairs without prompting (`--repair` is an alias) and enter maintenance under the [service-preservation and installation-drift rules](/cli/doctor/recovery#gateway-service-recovery). </Tab> <Tab title="--lint"> ```bash openclaw doctor --lint openclaw doctor --lint --json ``` Run structured health checks for CI or preflight automation. Read-only: no prompts, repairs, migrations, restarts, or state writes. </Tab> <Tab title="--fix --force"> ```bash openclaw doctor --fix --force ``` Apply aggressive config/state repairs too. Repair maintenance uses the same service-preservation and installation-drift rules; use `openclaw gateway install --force` from the intended installation to replace its launcher and managed environment. </Tab> <Tab title="--non-interactive"> ```bash openclaw doctor --non-interactive ``` Run without prompts, applying only safe migrations (config normalization + on-disk state moves). Skips restart/service/sandbox actions that need human confirmation. Legacy state migrations still run automatically when detected. Add `--fix` for all supported startup-blocking repairs without prompts, including workspace setup, session stores, exec approvals, and audit schema migrations. Explicit repair checks ownership before database snapshots; another live owner must stop before repair can proceed. Malformed or conflicting retained files require the manual recovery named in the error. When the shared database is already current, preparation leaves the running Gateway's worker environments available. Actual schema repairs retire the old database resources before later maintenance continues, including when repair cleanup fails. </Tab> <Tab title="--deep"> ```bash openclaw doctor --deep ``` Scan system services for extra gateway installs (launchd/systemd/schtasks). </Tab> </Tabs> With `OPENCLAW_GATEWAY_STARTUP_TRACE=1`, Doctor prints per-phase timings (`doctor.*` and `cli.bootstrap.*` lines) to stderr. To review changes before writing, open the config file first: ```bash cat ~/.openclaw/openclaw.json ``` ## Read-only lint mode `openclaw doctor --lint` is the automation-friendly sibling of `openclaw doctor --fix`. They share the same Doctor rule registry, but they do not select or act on rules in the same way: | Mode | Prompts | Writes config/state | Output | Use it for | | ------------------------ | --------- | ------------------------------------------ | ---------------------- | -------------------------------- | | `openclaw doctor` | yes | yes, safe migrations and confirmed repairs | friendly health report | guided checks and repairs | | `openclaw doctor --json` | no | no | JSON advisory report | machine-readable operator checks | | `openclaw doctor --fix` | sometimes | yes, with repair policy | friendly repair log | applying approved repairs | | `openclaw doctor --lint` | no | no | structured findings | CI, preflight, and review gates | Default `doctor --lint` runs the broad-safe automation profile: checks that are static, local, and useful in CI or preflight output. It skips opt-in checks that are advisory, environment-sensitive, live-service dependent, account/workspace inventory, or historical cleanup. Use `doctor --lint --all` when you want the full registered lint audit, including those opt-in checks, or `--only <id>` for a targeted check. `doctor --fix` does not use the lint default profile and does not accept `--all`. It runs Doctor's ordered repair path: modern health checks may provide an optional `repair()` implementation, and older areas still use their legacy Doctor repair flow. Some lint findings are intentionally diagnostic only, so a check appearing in `--lint --all` does not mean `--fix` will mutate that area. The contract separates `detect()` (reports findings) from `repair()` (reports changes/diffs/side effects), which keeps a path open for a future `doctor --fix --dry-run` without turning lint checks into mutation planners. Some built-in checks are default-disabled internally so they stay available to `--all`, `--only`, and Doctor repair flows without becoming part of the default `doctor --lint` automation profile. Finding severity is still emitted per finding (`info`, `warning`, or `error`); default selection is not a severity level. ```bash openclaw doctor --lint openclaw doctor --lint --severity-min warning openclaw doctor --lint --json openclaw doctor --lint --all openclaw doctor --lint --only core/doctor/gateway-config --json ``` JSON output fields: - `schemaVersion`: version of the machine-readable lint envelope; branch on this before parsing other fields - `ok`: whether any finding met the selected severity threshold - `checksRun` / `checksSkipped`: counts (skipped by profile, `--only`, or `--skip`) - `findings`: structured diagnostics with `checkId`, `severity`, `message`, and optional `path`, `line`, `column`, `ocPath`, `source`, `target`, `requirement`, `fixHint` Exit codes: | Code | Meaning | | ---- | -------------------------------------------------------- | | `0` | no findings at or above the selected threshold | | `1` | one or more findings met the selected threshold | | `2` | command/runtime failure before findings could be emitted | These threshold-based exit codes belong to explicit `--lint` mode, with or without `--json`. Bare `openclaw doctor --json` preserves ordinary Doctor's advisory exit `0` after producing its payload; machine consumers should read `ok` and `findings`. Fatal errors before output remain nonzero. During `openclaw update`, failure to remove Doctor's disposable lint snapshot is recorded as an update warning and does not block the update. Standalone `doctor --lint` still reports that cleanup failure as an error. The update keeps the checks' actual findings; cleanup warnings do not hide other failures. Flags: - `--severity-min info|warning|error` (default `warning`): controls both what prints and what causes a non-zero exit. - `--all`: runs every registered lint check, including opt-in checks excluded from the default automation set. - `--only <id>` (repeatable): run only the named check id(s); an unknown id is reported as an error finding. - `--skip <id>` (repeatable): exclude a check while keeping the rest of the run active. - `--severity-min`, `--all`, `--only`, and `--skip` require `--lint`. Bare `--json` is allowed for an advisory machine-readable report; `--fix` rejects it unless another machine mode owns the output.