Analytics Perspective (ANL)

Evaluates the delivery and process health of a project by analysing CI/CD automation, commit patterns, and working tree state. Assigns a score from 0.0 to 1.0 across five dimensions.

Dimensions

CI/CD — Is there automated validation, and is it meaningful?

• 1.0: pipeline present AND includes a test-run step AND is passing
• 0.5: pipeline present but no test step, or pipeline is failing
• 0.0: no pipeline at all
• To check CI status: gh run list --limit 1 --json conclusion --jq '.[0].conclusion'
• If gh is unavailable, mark as 'unknown' and score 0.5

Commit quality — Does the commit history signal healthy development?

• Examines the last 20 commits
• If >50% of commits are bug fixes, the score is reduced: 1 - fix_rate
• A high fix rate signals weak test coverage, recurring design debt, or a broken development cycle

Working tree — Is the branch clean?

• No uncommitted changes = 1.0
• Any uncommitted changes = 0.5 (work in flight, not an error, but a signal)

Commit message quality — Are commit messages informative?

• Examine last 20 commits: what fraction have vague messages ("fix", "WIP", "update", "changes")?
• Score: 1.0 if <10% vague, 0.5 if 10–40% vague, 0.0 if >40% vague
• Finding priority: low

Test automation — Are tests integrated into the delivery pipeline?

• Score: 1.0 if CI runs tests, 0.5 if tests exist but not in CI, 0.0 if no tests
• Finding priority: medium

Scoring

anl_score = (ci_score + commit_quality + clean + msg_quality + test_automation) / 5

Score interpretation:

• 0.9–1.0 Delivery pipeline is healthy and automated
• 0.7–0.9 Good CI/CD with some process noise
• 0.5–0.7 Missing automation or high fix-to-feature ratio
• 0.0–0.5 No CI/CD and commit history shows a reactive, fire-fighting pattern

Tasks Generated

Finding	Priority
No CI/CD pipeline	medium
CI pipeline present but missing test step	medium
Fix rate > 50% in last 20 commits	medium
Uncommitted changes in working tree	medium
>40% vague commit messages in last 20 commits	low
Tests exist but not run in CI	medium
No test suite found	medium

Agent Analysis Guide

When performing a manual ANL analysis:

1. Check .github/workflows/ — verify at least one workflow runs tests on push/PR; check whether the pipeline is currently passing. Use gh run list --limit 1 --json conclusion --jq '.[0].conclusion' for automated status check. If gh is unavailable, mark as 'unknown' and score 0.5.
2. Run git log -20 --oneline — classify each commit: first check for conventional-commit prefixes (fix:, bugfix:, hotfix: for fixes; feat: for features; chore:, refactor:, docs: for other categories); if no prefix is present, apply keyword fallback (look for fix, feature, add, remove, update, refactor in message); as last resort, use subjective inference from the message content.
3. Calculate fix rate — if >50%, identify which modules are generating the most fixes
4. Run git status — note any uncommitted changes and their age (stash date)
5. Assess commit message quality — count messages that are vague ("fix", "WIP", "update", "changes", "stuff"); score as 1.0/<10%, 0.5/10–40%, 0.0/>40%
6. Check test automation — does the project have a test script in package.json / Makefile / equivalent, and does the CI workflow invoke it?
7. Look for patterns: are fixes always in the same module? That module likely needs a design review

Output: a delivery health assessment with the fix rate, commit message quality ratio, test automation status, and specific automation gaps.