release-manager（发布经理）

Release Manager

The agent automates release management by parsing conventional commits into structured changelogs, determining semantic version bumps, and assessing release readiness with checklists, rollback runbooks, and stakeholder communication plans.

Quick Start

# Generate changelog from conventional commits
git log --oneline v1.0.0..HEAD | python changelog_generator.py --version 1.1.0 --format both

# Determine version bump from commit history
git log --oneline v1.0.0..HEAD | python version_bumper.py --current-version 1.0.0 --analysis

# Assess release readiness
python release_planner.py --input release-plan.json --include-checklist --include-rollback

Core Workflows

Workflow 1: Generate Changelog and Version Bump

1. Collect commits since last tag: git log --oneline v1.0.0..HEAD
2. Pipe to changelog_generator.py to produce a Keep-a-Changelog-format CHANGELOG
3. Pipe to version_bumper.py to determine MAJOR/MINOR/PATCH bump from commit types
4. Review changelog grouping (Added, Fixed, Changed, Breaking Changes)
5. Validation checkpoint: All feat commits appear under Added; all fix under Fixed; breaking changes highlighted

git log --oneline v1.2.0..HEAD | python changelog_generator.py \
  --version 1.3.0 --date 2026-03-21 --base-url https://github.com/org/repo --summary

Workflow 2: Assess Release Readiness

1. Prepare release plan JSON with features, quality gates, stakeholders, and target date
2. Run release_planner.py with checklist, communication, and rollback flags
3. Review blocking issues and readiness score
4. Address blockers (missing approvals, failed gates, overdue items)
5. Validation checkpoint: Readiness score >80%; zero blocking issues; rollback runbook generated with time estimates

python release_planner.py --input release-plan.json \
  --output-format json --include-checklist --include-communication --include-rollback

Workflow 3: Hotfix Release

1. Create hotfix branch from last stable tag
2. Apply minimal fix and run version_bumper.py with --prerelease rc
3. Generate changelog entry for the fix
4. Assess readiness with expedited checklist
5. Validation checkpoint: Fix addresses root cause only; rollback procedure tested; stakeholders notified

Version Bump Rules

Pre-release progression: alpha.N -> beta.1 -> rc.1 -> stable release.

Rollback Triggers

• Error rate: >2x baseline within 30 minutes
• Latency: >50% P95 increase
• Feature failures: Core functionality broken
• Security incident: Vulnerability exploited
• Data corruption: Database integrity compromised

Anti-Patterns

• Monolithic releases -- large, infrequent releases with high blast radius; prefer small, frequent releases
• Manual deployments -- error-prone and inconsistent; automate every step that can be automated
• No rollback plan -- every release must have a tested rollback procedure before going live
• Skipping quality gates -- deploying without test coverage, security scan, or dependency audit
• Last-minute changes -- code freeze exists for a reason; changes after freeze need explicit approval
• Non-conventional commits -- free-form commit messages break changelog generation and version bumping
• Environment drift -- staging must mirror production; drift causes false confidence in testing

Troubleshooting

Success Criteria

• Changelog accuracy: 100% of conventional commits are correctly categorized (feat to Added, fix to Fixed, etc.) with zero miscategorized entries
• Version bump correctness: Recommended version matches SemVer rules in all cases -- breaking changes produce MAJOR, features produce MINOR, fixes produce PATCH
• Readiness assessment coverage: All blocking issues (missing approvals, failed quality gates, overdue timelines) are surfaced with zero false negatives
• Release cycle time reduction: Teams using the planner reduce release preparation time by 40% or more compared to manual checklist tracking
• Rollback preparedness: Every release assessed by the planner has a complete, actionable rollback runbook with time estimates and verification steps
• Stakeholder communication: Communication plans cover all identified stakeholders with appropriate timing (T-48h external, T-24h internal, T+1h post-deploy)
• Tool integration time: New teams can configure and run all three scripts against their repository within 30 minutes of initial setup

Scope & Limitations

This skill covers:

• Parsing conventional commits and generating structured changelogs in Markdown and JSON formats
• Determining semantic version bumps (major/minor/patch) with pre-release support (alpha, beta, rc)
• Assessing release readiness across features, quality gates, approvals, and timelines
• Generating rollback runbooks, communication plans, and release checklists from structured input

This skill does NOT cover:

• Actual CI/CD pipeline execution or deployment automation (see engineering/ci-cd-pipeline-generator)
• Live monitoring, alerting, or incident response during deployments (see engineering/monitoring-alerting-setup)
• Code review processes or pull request management (see engineering/code-review-automation)
• Infrastructure provisioning, container orchestration, or environment management (see engineering/infrastructure-as-code)

Integration Points

Tool Reference

changelog_generator.py

Purpose: Parses git log output in conventional commit format and generates structured changelogs. Groups commits by type (Added, Fixed, Changed, etc.), extracts scope and issue references, and highlights breaking changes.

Usage:

git log --oneline v1.0.0..HEAD | python changelog_generator.py
python changelog_generator.py --input commits.txt --version 2.0.0 --format json
cat commits.json | python changelog_generator.py --input-format json --summary

Flags/Parameters:

Example:

git log --oneline v1.2.0..HEAD | python changelog_generator.py \
  --version 1.3.0 \
  --date 2026-03-21 \
  --base-url https://github.com/myorg/myapp \
  --format both \
  --summary

Output Formats:

• markdown: Keep a Changelog format with sections for Breaking Changes, Added, Changed, Deprecated, Removed, Fixed, Security. Commits grouped by scope within each section.
• json: Structured object with version, date, summary (counts by type, by author, scopes, issue references), and categories (arrays of commit objects per category).
• both: Markdown changelog followed by JSON output, each with a heading separator.

version_bumper.py

Purpose: Analyzes conventional commits since the last tag to determine the correct semantic version bump (major/minor/patch). Supports pre-release versions (alpha, beta, rc) and generates version bump commands for npm, Python, Rust, Git, and Docker.

Usage:

git log --oneline v1.2.0..HEAD | python version_bumper.py --current-version 1.2.0
python version_bumper.py -c 2.0.0-beta.3 -i commits.json --input-format json --prerelease rc
git log --oneline v1.0.0..HEAD | python version_bumper.py -c 1.0.0 -f json --analysis --include-commands

Flags/Parameters:

Example:

git log --oneline v2.1.0..HEAD | python version_bumper.py \
  --current-version 2.1.0 \
  --output-format json \
  --analysis \
  --include-commands \
  --include-files

Output Formats:

• text: Human-readable summary with current version, recommended version, bump type, and optional analysis/commands.
• json: Structured object with current_version, recommended_version, bump_type, and optional analysis, commands, and file_updates fields.
• commands: Shell-ready version bump commands organized by platform (npm, Python, Rust, Git, Docker).

release_planner.py

Purpose: Takes a release plan JSON (features, quality gates, stakeholders, target date) and assesses release readiness. Generates a readiness report with scoring, a release checklist, a stakeholder communication plan with message templates, and a rollback runbook.

Usage:

python release_planner.py --input release-plan.json
python release_planner.py -i plan.json -f json --include-checklist --include-rollback
python release_planner.py -i plan.json -f markdown --include-checklist --include-communication --include-rollback

Flags/Parameters:

Example:

python release_planner.py \
  --input release-plan.json \
  --output-format json \
  --output readiness-report.json \
  --include-checklist \
  --include-communication \
  --include-rollback \
  --min-coverage 85.0

Input JSON Structure: The input file expects a JSON object with keys: release_name, version, target_date (ISO format), features (array of feature objects with id, title, type, status, risk_level, approvals, etc.), quality_gates (optional array), stakeholders (optional array), and rollback_steps (optional array). When quality_gates or rollback_steps are omitted, sensible defaults are generated automatically.

Output Formats:

• text: Plain text report with status, readiness score, blocking issues, warnings, recommendations, feature summary, and quality gate summary.
• markdown: Formatted Markdown report with headings, status icons, and structured feature/checklist sections.
• json: Complete structured object with assessment, checklist, communication_plan, and rollback_runbook fields (null when not requested).