devops

DevOps

Overview

Responsible for infrastructure automation, CI/CD pipeline health, and GitHub project tracking. Primary mission: enable autonomous sprint visibility through GitHub Projects v2, automated reporting, and deployment reliability.

When to Use

• Setting up or modifying CI/CD workflows (GitHub Actions)
• Configuring GitHub Projects v2 (boards, views, custom fields)
• Automating sprint visibility (burndown, velocity, status sync)
• Debugging deployment pipelines or infrastructure issues
• Managing GitHub token scopes and permissions

When NOT to Use

• For sprint ceremony process — that's scrum-master
• For code quality standards — that's python-quality
• For pre-commit hook configuration — that's quality-gates
• For agent assignment — that's agent-delegation

Core Process

CI/CD Pipeline Architecture

Push/PR to main
  ↓
GitHub Actions: quality-tests.yml
  ├── ruff format --check
  ├── ruff check
  ├── pytest (Python 3.11, 3.12 matrix)
  └── security scan (bandit)
  ↓
Branch protection: require CI pass + PR approval
  ↓
Merge to main
  ↓
Content pipeline (content-pipeline.yml) — on schedule or dispatch
  ├── Topic discovery
  ├── Editorial review
  ├── Content generation
  ├── Quality gates
  └── Blog deployment

GitHub Projects v2 Setup

Required token scopes:

gh auth refresh -s project
gh auth status  # verify: project, read:org, repo, workflow

Board creation and configuration:

gh project create "Sprint N: Title" --owner oviney --format "Board"

Custom fields: Sprint (single select), Story Points (number), Priority (single select), Owner (single select).

Views: Kanban (status lanes), Table (sortable), Roadmap (timeline).

Sprint Tracking Automation

Component	Purpose	Location
sprint_tracker.json	Sprint state, velocity, completion	data/skills_state/
github_sprint_sync.py	Bidirectional SPRINT.md <-> GitHub	scripts/
sprint-sync.yml	Auto-update on issue close	.github/workflows/
sprint-discipline.yml	Commit message enforcement	.github/workflows/

Deployment Pipeline

Content pipeline stages:

1. Topic Scout discovers topics
2. Editorial Board votes
3. Stage 3 generates article (research, graphics, writing)
4. Stage 4 reviews (quality gate, schema validation, publication validation)
5. Deploy to blog repo if quality gate passes

Key configuration:

• BLOG_REPO_TOKEN — auth for blog repo push
• OPENAI_API_KEY, ANTHROPIC_API_KEY — LLM API access
• SERPER_API_KEY — Google Scholar/web search (optional)

Monitoring and Reporting

Metric	Source	Automation
Sprint velocity	sprint_tracker.json	generate_sprint_badge.py
Test count	pytest discovery	generate_tests_badge.py
Quality score	quality_dashboard.py	quality_score.json
CI pass rate	GitHub Actions API	gh run list
Content pipeline success	content-pipeline.yml logs	Manual review

Common Rationalizations

Rationalization	Reality
"Manual deployment is fine for now"	Manual steps are forgotten steps — automate the pipeline and catch failures in CI
"We don't need GitHub Projects, issues are enough"	Issues track work; Projects track flow — burndown and velocity need the structured views
"Token scope refresh is risky"	gh auth refresh -s project adds scope without invalidating existing permissions
"We'll monitor CI manually"	Unmonitored pipelines fail silently — badge automation makes failures visible in README
"The pipeline works locally, CI will be fine"	Local has different env vars, Python versions, and dependencies — CI is the source of truth

Red Flags

• CI pipeline failing on main branch (broken build)
• GitHub token missing required scopes (project, repo, workflow)
• Sprint tracker and GitHub issues out of sync
• Content pipeline running without quality gate checks
• Badge JSON files stale (not reflecting actual metrics)
• Deployment to blog repo without BLOG_REPO_TOKEN verification
• Workflow files modified without testing in a PR first

Verification

• CI runs on every PR to main — evidence: gh run list shows recent runs for PRs
• Quality tests pass: ruff, pytest, security scan — evidence: green CI checks on latest PR
• GitHub Projects board exists for current sprint — evidence: gh project list shows board
• Sprint tracker in sync with GitHub — evidence: github_sprint_sync.py --show-github-status shows no discrepancies
• Badges reflect current metrics — evidence: validate_badges.py passes
• Content pipeline deploys successfully — evidence: latest content-pipeline.yml run is green

Integration Points

• .github/workflows/quality-tests.yml — CI quality checks
• .github/workflows/content-pipeline.yml — article generation and deployment
• .github/workflows/sprint-sync.yml — sprint status auto-update
• scripts/github_sprint_sync.py — bidirectional sprint sync
• scripts/validate_badges.py ��� badge accuracy validation
• data/skills_state/sprint_tracker.json — sprint state tracking