red-doctor

Diagnoses repository adoption of the RedSkills engineering stack and applies gated canonical fixes.

20|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reddb-io/red-skills --skill red-doctor-reddb-io
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: red-doctor
Source: https://github.com/reddb-io/red-skills/tree/main/plugins/dev/skills/engineering/red-doctor
Command: npx skills add https://github.com/reddb-io/red-skills --skill red-doctor-reddb-io

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Repositories drift from canonical engineering conventions over time: labels diverge, config keys land in legacy namespaces, MCP wiring goes stale, cached plugin runtimes go missing, and dependency metadata desynchronizes. Manually auditing dozens of adoption checks across a repo is slow and error-prone, and unsafe auto-fixes can cause more damage than the drift itself. ## Core Features & Use Cases - Read-only diagnostic scorecard: Runs 30 checks covering label conformance, AGENTS/CLAUDE parity, config namespacing, statusline drift, MCP wiring, version coherence, workflow naming, runtime distribution, dependency-edge audits, host toolchain versions, and more — without mutating anything by default. - Gated --fix reconciliation: Applies each finding's canonical fix, batching safe idempotent repairs and requiring explicit per-item confirmation for hard-to-reverse changes like label renames, config migrations, and cache re-fetches. - Delegation to single-writer tools: Never hand-edits what another tool owns — version mismatches go to the release tool, setup gaps to /red-setup, and ticket authoring to /triage. - Use Case: Before launching a large autonomous /afk work drain, run the doctor to verify the repo's labels, hooks, runtime bundles, and toolchain all conform, then apply fixes with confirmation gates. ## Quick Start Ask the agent to run red doctor against this repository and report the adoption scorecard, optionally with --fix to apply the canonical fixes.

Frequently Asked Questions about red-doctor

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

FAQPage Schema
How do I check if my repo follows RedSkills conventions?▼

Run red doctor against the repository to get a read-only scorecard of adoption findings. It audits labels, config namespacing, MCP wiring, runtime bundles, workflow naming, and toolchain versions, tagging each gap with its canonical fix-home.

How do I automatically fix RedSkills adoption drift?▼

Invoke red doctor with the --fix flag. Safe idempotent fixes apply in a batch with receipts, while hard-to-reverse changes like label renames, config migrations, and cache re-fetches require explicit per-item confirmation before running.

Does red doctor modify my repository by default?▼

No, the default pass is strictly read-only with no label edits, file writes, or installs. Only the explicit --fix flag enables mutations, and even then every hard-to-reverse change is confirmed individually.

What is the difference between red doctor and red-setup?▼

red-setup is the one-time adoption installer that creates a repo's .red/ tree and initial wiring. red doctor is the recurring counterpart that audits ongoing conformance and heals drift, delegating creation tasks back to red-setup.

Why does red doctor delegate some fixes instead of applying them?▼

Fixes owned by single-writer tools — version manifests, marketplace registrations, ticket authoring — are delegated to avoid two writers corrupting the same state. The doctor names the owning tool for each finding rather than patching around it.