Orchestrate 6-phase changelog generation workflow with AI synthesis, multi-source data fetching (GitHub/Slack/Git), quality validation, and automated PR creation. Use when automating release notes, weekly changelogs, or documentation updates. Trigger with "generate changelog", "weekly changelog", or "automate release notes".
Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Orchestrate 6-phase changelog generation workflow with AI synthesis, multi-source data fetching (GitHub/Slack/Git), quality validation, and automated PR creation. Use when automating release notes, weekly changelogs, or documentation updates. Trigger with "generate changelog", "weekly changelog", or "automate release notes".
allowed-tools
Read,Write,Bash(python:*)
version
1.0.0
author
Jeremy Longshore <jeremy@intentsolutions.io>
license
MIT
compatibility
Designed for Claude Code; requires gh CLI for GitHub source and slack-sdk + NIXTLA_SLACK_TOKEN for optional Slack source
Orchestrate a 6-phase workflow to automate changelog generation: fetch data from multiple sources, synthesize narrative with AI, validate quality, and create pull requests.
Overview
This skill coordinates changelog automation through progressive disclosure:
Phase 1: Initialize & fetch data from GitHub/Slack/Git
Phase 2: AI synthesis (Writer Agent groups changes, generates narrative)
Phase 3: Template formatting with frontmatter validation
Phase 4: Quality review with two-layer gate (deterministic + editorial)
Phase 5: PR creation with changelog file
Phase 6: User handoff with summary and next steps
Integrates with MCP server (changelog_mcp.py) for deterministic operations while handling editorial work (synthesis, tone, quality judgment) directly.
Prerequisites
Environment:
Python 3.10+ installed
MCP server dependencies installed: pip install -r scripts/requirements.txt
.changelog-config.json configured in project root
Required Tokens:
GitHub: GITHUB_TOKEN environment variable with repo:read + repo:write scopes
Slack (optional): SLACK_TOKEN environment variable with channels:history scope
Files:
Config file: .changelog-config.json (see config/.changelog-config.example.json)
Template: Markdown file with frontmatter (see assets/templates/default-changelog.md)
Instructions
Phase 1: Initialize & Fetch Data
Load Configuration
Call MCP tool get_changelog_config (no arguments for default .changelog-config.json)
If error: Display error message with suggestion (e.g., "Run /changelog-validate")
For /changelog-custom: Use provided start_date and end_date parameters
Validate dates are in ISO 8601 format (YYYY-MM-DD)
Ensure start_date < end_date
Fetch Data from All Sources
For each source in config["sources"]:
Call MCP tool fetch_changelog_data with:
source_type: From config (github/slack/git)
start_date: Calculated or provided date
end_date: Calculated or provided date
config: Source-specific config from file
Collect items from response data["items"]
Aggregate all items into unified dataset
Sort by timestamp (oldest to newest)
Display Fetch Summary
Total items fetched
Breakdown by source (e.g., "GitHub: 12 PRs, Slack: 5 messages")
Date range covered
Phase 2: Writer Agent - AI Synthesis
Role: You are now the Writer Agent. Your job is to transform raw data into user-friendly changelog content.
Group Items by Type
Features: Items with labels/types: enhancement, feature, new
Fixes: Items with labels/types: bug, fix, bugfix
Breaking Changes: Items with labels: breaking, breaking-change
Other: Everything else
Generate Narrative Summary
Write 2-3 sentences summarizing the overall theme of changes
Focus on what matters to users (not internal details)
Example: "This week focused on improving performance and user experience. Key highlights include a new dark mode toggle and faster page load times. Several critical bugs were fixed in the checkout flow."
Role: You are the Orchestrator Agent again. Present results to the user.
Display Success Summary
🎉 Changelog PR created successfully!
📋 Summary:
- Date range: {start_date} to {end_date}
- Changes: {count} items ({source breakdown})
- Quality score: {score}/100
- PR: {pr_url}
Next steps:
1. Review PR: Click link above
2. Merge when ready: Changelog will be added to repo
3. Iterate: Run /changelog-custom for different date ranges
Provide Next Actions
Review PR (link)
Merge when satisfied
Run /changelog-custom for different date range
Adjust quality threshold in config if needed
Save Reproducibility Bundle (Optional)
Create run_manifest.json with:
Execution timestamp
MCP server version
Data source versions
Item count per source
Quality scores
PR metadata
Store in project root or .claude/ directory
End Session
Thank user
Remind them of validation command: /changelog-validate
Output
Primary Artifact: GitHub Pull Request with changelog file
Secondary Artifacts:
Changelog markdown file (e.g., CHANGELOG.md)
Run manifest JSON (optional, for reproducibility)
Console Output: Phase-by-phase progress with status emojis
