| name | health-check |
| description | Run a full Plan Forge health diagnostic — environment, setup validation, and completeness scan. Use to verify your forge is properly configured and code is clean. |
| argument-hint | [optional: specific check to focus on — 'environment', 'setup', or 'sweep'] |
| tools | ["forge_smith","forge_validate","forge_sweep"] |
Health Check Skill
Trigger
"Check my setup" / "Is my forge healthy?" / "Run health check" / "Diagnose my project"
Steps
1. Environment Diagnostics
Use the forge_smith MCP tool to inspect the forge — diagnose environment, VS Code config, setup health, version currency, and common problems.
Review the output for:
- Required tools installed (git, VS Code, PowerShell/Bash, gh CLI)
- VS Code settings for Copilot agent mode
- .forge.json validity and version currency
- MCP server status
- Orchestrator readiness
If forge_smith reports failures: List each failure with its FIX recommendation. Continue to Step 2 regardless.
2. Setup Validation
Use the forge_validate MCP tool (or run pforge check) to verify that all required Plan Forge files exist, file counts match preset expectations, and no unresolved placeholders remain.
Review the output for:
- Required files present (instructions, agents, prompts, skills)
- File counts match preset expectations
- No unresolved
Falsify placeholders
- AGENTS.md and copilot-instructions.md configured
If validate reports missing files: Flag as CRITICAL — setup may need to be re-run.
3. Completeness Scan
Use the forge_sweep MCP tool to scan code files for TODO, FIXME, HACK, stub, placeholder, and mock data markers.
Review the output for:
- Count of deferred-work markers
- Location of each marker
- Whether markers are in production code vs. test/config files
If sweep finds markers in production code: Flag as WARNING — these should be resolved before the Review Gate (Step 5).
4. Report
Plan Forge Health Check:
Environment: N passed, N failed, N warnings
Setup: N passed, N failed
Sweep: N deferred-work markers found
Critical Issues: (list any)
Warnings: (list any)
Recommendations: (list any)
Overall: PASS / FAIL
A PASS requires:
- Zero environment failures (warnings OK)
- Zero setup failures
- Zero sweep markers in production code (markers in tests/config are OK)
Safety Rules
- This skill is READ-ONLY — do NOT modify any files
- Report findings clearly with actionable fix recommendations
- Distinguish between critical issues (must fix) and warnings (should fix)
Temper Guards
| Shortcut | Why It Breaks |
|---|
| "Smith passed, so everything is fine" | Smith checks environment and config, not code quality or plan state. A passing smith doesn't mean your sweep is clean or your plans are valid. |
| "The sweep found markers but they're just TODOs in tests" | Sweep markers in test files are INFO-level, but markers in production code block gate passage. Report them separately — don't dismiss the whole sweep. |
| "Setup validation isn't needed — I ran setup recently" | File counts drift when files are manually deleted or instructions updated. Always validate after any change to the .github/ directory. |
Warning Signs
- Health check skipped one or more diagnostic steps (environment, setup, sweep)
- Sweep markers in production code reported as OK
- Setup validation failures dismissed without running
pforge check or re-running setup
- Report shows PASS when critical issues were found but downgraded
Exit Proof
After completing this skill, confirm:
Persistent Memory (if OpenBrain is configured)
- Before checking:
search_thoughts("setup issue", project: "Falsify", created_by: "copilot-vscode", type: "bug") — load prior setup issues and recurring health check failures
- After check:
capture_thought("Health check: <summary of findings>", project: "Falsify", created_by: "copilot-vscode", source: "skill-health-check") — persist environment issues for future diagnostics