Speckit Doctor

User Input

$ARGUMENTS

Outline

Run a comprehensive health check on the spec-kit project, validating that all required components are present and correctly configured.

Step 1: Find the project root

Locate the .specify/ directory. If not found, error: "No spec-kit project found. Run specify init first."

Step 2: Check templates

Verify these template files exist in .specify/templates/ and are non-empty:

• spec-template.md
• plan-template.md
• tasks-template.md
• constitution-template.md
• checklist-template.md

For each template:

• File exists and non-empty: PASS
• File exists but empty: WARN ("template is empty")
• File missing: FAIL ("template not found")

Step 3: Check agent configuration

Read .specify/init-options.json to determine the configured AI agent.

If no init-options.json exists: WARN ("no AI agent configured - run specify init to set up")

If an agent is configured (e.g., "ai_assistant": "claude"):

1. Check that the agent's directory exists (e.g., .claude/)
2. Check that command files are registered (e.g., .claude/commands/speckit.*.md)
3. Count registered commands

• Agent dir exists with commands: PASS
• Agent dir missing: FAIL ("agent directory not found")
• Agent dir exists but no commands: WARN ("no command files registered")

Step 4: Check the Python runner and helpers

Look for the current SpecKit Pro runner package in either location:

• Source checkout, relative to the project root: speckit-pro/speckit_pro_runner/
• Installed payload, relative to the resolved plugin root: speckit_pro_runner/

If a runner package is present, verify these files exist and are non-empty:

• __main__.py
• runtime.py
• helpers/registry.py
• gates/registry.py
• speckit-pro-runner.manifest.json
• speckit-pro-runner.sha256

Read the manifest with Python standard-library json and confirm it is a JSON object.

• Runner package and all required files are valid: PASS
• Runner package exists but a required file is missing, empty, or invalid: FAIL
• Neither runner location exists: WARN ("SpecKit Pro Python runner not found - skipping plugin runtime layout checks")

The vendored .specify/scripts/** compatibility helpers are outside this SpecKit Pro runtime check. Leave them unchanged and do not infer a plugin runtime prerequisite from their presence.

Step 5: Check constitution

Look for constitution.md or .specify/memory/constitution.md in the project root.

• Exists and has content (>10 words): PASS (show word count)
• Exists but empty: WARN ("constitution is empty")
• Not found: WARN ("no constitution found - consider creating one to guide AI decisions")

Step 6: Check features

Scan specs/ for numbered feature directories. For each feature:

Check for required artifacts: spec.md, plan.md, tasks.md

• All three present: PASS
• Some missing: WARN (list which are missing, e.g., "spec ✓ plan ✗ tasks ✗")

Step 7: Report

Output a summary table:

Project Health Check
====================

Templates:     5/5 PASS
Agent Config:  PASS (Claude Code, 8 commands registered)
Python Runner: PASS (source checkout; helpers, gates, metadata)
Constitution:  PASS (245 words)

Features:
  001-auth:      spec ✓  plan ✓  tasks ✓  PASS
  002-dashboard: spec ✓  plan ✗  tasks ✗  WARN (needs /speckit.plan)

Overall: 4 PASS, 1 WARN, 0 FAIL

If any FAIL results exist, suggest specific remediation steps.