doctor

Audits Beislið workflow configuration and probes each capability against the current host session.

10|Updated May 12, 2026
One-click install
npx skills add https://github.com/sandsower/beislid --skill doctor-sandsower
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: doctor
Source: https://github.com/sandsower/beislid/tree/main/skills/doctor
Command: npx skills add https://github.com/sandsower/beislid --skill doctor-sandsower

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Teams using Beislið configure their agent workflow in .beislid/workflow.md, but misconfigured gates, missing binaries, or stale probe results silently break orchestration. Doctor audits the entire config end-to-end, probes each capability, and reports gaps in conversational prose without modifying anything. ## Core Features & Use Cases - Config validation: Parses every beislid:<key> fenced block in workflow.md, validating gates, gate sets, lifecycle actions, hooks, review policy, action policy, model routing, and more against the format spec. - Capability probing: Runs binary probes (e.g., command -v, nopal info --json) for executable capabilities and records results in a JSON probe cache that orchestrators consume for lazy-probing decisions. - Drift detection with caching: Hashes workflow.md and skips re-probing when the cache is fresh; --refresh forces a full re-probe. - Use Case: After editing workflow.md to add a new pre-PR gate, run doctor to confirm the gate command exists, the YAML parses, and the cache orchestrators rely on reflects the change. ## Quick Start Ask the agent to run doctor to audit my Beislið workflow configuration and probe every capability.

Frequently Asked Questions about doctor

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check if my Beislið workflow.md config is valid?▼

Run the doctor skill, which reads .beislid/workflow.md, validates every beislid:<key> fenced block against the format spec, and reports parse errors, unknown keys, and invalid values with line numbers in conversational prose.

How do I force doctor to re-probe capabilities instead of using the cache?▼

Pass --refresh when invoking doctor. This skips the freshness check entirely and runs a full re-probe, then overwrites the cached JSON in the Beislið state directory.

Does doctor modify my workflow.md or install anything?▼

No. Doctor is read-only on the repository and write-only on its probe cache. It never edits workflow.md, installs or removes skills, posts comments, creates worktrees, or invokes other orchestrators.

Why does doctor fail with a git repository error?▼

Doctor requires a git repository with at least one commit because it derives the repo identity hash from the root commit and locates config at the repo root. Run git init and make an initial commit, then re-run doctor.

What happens when doctor finds a duplicate or unknown config key?▼

For duplicate beislid:<key> blocks, the first occurrence wins and doctor warns with the line number of the duplicate. Unknown keys are skipped with a note suggesting you check the format spec, and neither case stops the rest of the audit.

Where does doctor store its probe results?▼

Probe results are written as JSON to ${BEISLID_STATE_DIR:-$HOME/.local/state/beislid}/probes/<repo_hash>.json. The cache is always written, including failures, so orchestrators have full state for lazy-probing decisions.