Creates an example workflow job that analyzes recently merged PRs and updates AGENTS.md or copilot-specific instructions accordingly, with state tracking to prevent duplicate processing across runs.

Implementation

New standard job: pr_docs_update

• Three-step workflow: fetch → analyze → update
• State tracking via .deepwork/pr_docs_state.json to skip processed PRs
• Configurable PR count parameter for historical depth control
• Categorizes changes as general (AGENTS.md) or copilot-specific

Job structure:

steps:
  - fetch_prs: Retrieves merged PRs, filters against state file
  - analyze_prs: Identifies documentation-relevant changes
  - update_docs: Applies updates, maintains state file

Files added:

• src/deepwork/standard_jobs/pr_docs_update/ (source of truth)
• .deepwork/jobs/pr_docs_update/ (working copy)
• Includes job.yml, step instructions, AGENTS.md, USAGE.md

Usage pattern:

# Initial run
/pr_docs_update.fetch_prs pr_count=10
/pr_docs_update.analyze_prs
/pr_docs_update.update_docs

# Subsequent runs automatically skip processed PRs
/pr_docs_update.fetch_prs pr_count=20  # Go deeper into history

State file enables incremental processing - each run only processes new PRs, supports arbitrary lookback by adjusting pr_count.

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.