status

Project Confidence Dashboard

Gather all quality signals for the current project and present a scored dashboard. This gives Spencer an instant read on whether the code is ship-ready.

---

How to Run

Execute all checks, calculate the score, and present the dashboard. Run checks in parallel where possible.

Step 1: Detect Project Context

# Find project root
# Look for package.json, pyproject.toml, Cargo.toml, go.mod, .git

# Detect project name from:
# 1. PROJECT_STATE.md header
# 2. package.json name field
# 3. Directory name as fallback

# Detect project type:
# Has VERIFICATION.md or database config? → data-backed
# Has package.json with react/next/vue? → frontend
# Has pyproject.toml or setup.py? → Python
# Has Cargo.toml? → Rust
# Has go.mod? → Go

Step 2: Gather Signals

Run these checks. For each, capture the result and calculate the score.

Signal 1: Tests (35 points base)

# Python
pytest --tb=no -q 2>&1 | tail -5
# → Parse "X passed, Y failed, Z skipped"

# JavaScript/TypeScript
npx jest --silent 2>&1 | tail -5
# or: npx vitest run --reporter=verbose 2>&1 | tail -5

# Go
go test ./... 2>&1 | tail -10

# Rust
cargo test --quiet 2>&1 | tail -5

Scoring:

• 100% pass → 35 points
• 90-99% pass → 31 points
• 80-89% pass → 28 points
• 70-79% pass → 20 points
• <70% pass → 0 points
• No tests found → 10 points (benefit of doubt, but flag it)

Signal 2: Lint (15 points base)

# Run linter only on files changed since last commit
git diff --name-only HEAD

# Then lint each changed file with the appropriate linter
# Python: pylint --errors-only
# JS/TS: npx eslint (if config exists)
# Shell: shellcheck

Scoring:

• 0 errors, 0 warnings → 15 points
• 0 errors, some warnings → 10 points
• Any errors → 0 points
• No linter available → 12 points (neutral)

Signal 3: Data Verification (20 points base)

Only for projects with VERIFICATION.md.

# Check if VERIFICATION.md exists
# If yes, run verification queries (check_all.sql or check_all.py)
# Parse PASS/FAIL counts
# Also count FILL_IN placeholders (not yet verified)

Scoring:

• All tests PASS, no FILL_IN → 20 points
• All tests PASS, some FILL_IN → 15 points

• 80% PASS → 12 points

• <80% PASS → 5 points
• Not run or no VERIFICATION.md → 0 points (redistributed)

Redistribution when no database: Add 10 to Tests, 5 to Lint, 5 to Git.

Signal 4: Smoke Test (20 points base)

# Check if smoke_test.sh exists in project root
# If yes, run it and parse PASS/FAIL output

Scoring:

• All checks pass → 20 points
• Any fail → 0 points
• No smoke_test.sh → 5 points (benefit of doubt, but flag it)

Redistribution when no smoke test: Add 10 to Tests, 5 to Lint, 5 to Git.

Signal 5: Git Cleanliness (10 points base)

git status --porcelain | wc -l
# Count uncommitted changes

Scoring:

• 0 uncommitted code files → 10 points
• 1-4 uncommitted → 5 points
• 5+ uncommitted → 0 points

---

Step 3: Calculate & Present

Add up all signal scores. Present the dashboard:

## Project Status: {project_name}
Run: {timestamp} | Branch: {branch}

### Confidence: {score}/100 {progress_bar}

| Signal            | Status          | Details                     |
|-------------------|-----------------|-----------------------------|
| Tests             | {X/Y passing}   | {runner}, {skipped} skipped  |
| Lint              | {Clean/Warnings/Errors} | {error_count} errors, {warn_count} warnings |
| Data Verification | {X/Y passing}   | {fill_in} FILL_IN remaining |
| Smoke Test        | {All pass/X fail/N/A} | {check_count} checks   |
| Git               | {N uncommitted} | {file list or "clean"}      |

### Unresolved
{List any:}
- Failing tests with file:line
- TODO/FIXME/HACK in changed files (grep changed files)
- FILL_IN entries in VERIFICATION.md
- Pending tasks from task list

### Recommendation
{Based on score:}
- 90-100: "Ship it. All signals green."
- 70-89: "Almost there. Address the items above before completing."
- 50-69: "Significant gaps. Fix failing tests and verification before proceeding."
- <50: "Not ready. Major issues need attention."

Progress bar format: Use block characters proportional to score. Example: 87/100 █████████░

---

Rules

• Run ALL checks every time — don't skip signals
• If a check fails to run (tool not found, timeout), score it as 0 and note why
• Always show absolute counts (47/47), not just percentages
• The Unresolved section should be actionable — specific files and line numbers
• This skill is read-only — it reports, it doesn't fix