Add PR-triggered chapter review workflow and agent instruction set for PDF↔notebook quality gates

[Copilot]

This PR introduces a PR-driven multi-agent workflow for chapter development in NLPDE/, centered on strict chapter PDF ↔ notebook pairing and iterative agent feedback before human review. It also defines reusable agent instructions and shared-context guidance for consistent chapter evaluation.

• Workflow orchestration (.github/workflows/chapter-pr-agent-cycle.yml)

  • Triggers on PR updates touching chapter content or agent/workflow config.
  • Enforces exactly one changed chapter PDF and one changed notebook per PR, with identical base names.
  • Handles config-only updates gracefully by skipping chapter review jobs instead of failing.
  • Stages review flow into: content coverage review, interactivity review, then author revision-loop guidance.

• Agent contracts (.github/agent-instructions/*)

  • content-coverage-agent.md: requires chapter-to-notebook concept mapping; flags missing/insufficient coverage (not relevant extras).
  • interactivity-agent.md: prioritizes high-impact visualization/interactivity opportunities tied to chapter learning goals.
  • chapter-author-agent.md: defines iterative author behavior—submit chapter PR, incorporate agent feedback, repeat until both agent concerns are resolved.

• Process documentation (docs/agent-workflow.md)

  • Documents end-to-end PR loop and review responsibilities.
  • Specifies 1:1 filename rule semantics (exact pre-extension match).
  • Recommends one-time PDF preprocessing + shared vector store retrieval for multi-agent consistency and citation quality.

# Chapter pairing gate (excerpt)
if [[ ${#PDF_FILES[@]} -ne 1 || ${#IPYNB_FILES[@]} -ne 1 ]]; then
  echo "Expected exactly one chapter PDF and one notebook update per PR."
  exit 1
fi

PDF_BASE="$(basename "$PDF" .pdf)"
IPYNB_BASE="$(basename "$IPYNB" .ipynb)"
if [[ "$PDF_BASE" != "$IPYNB_BASE" ]]; then
  echo "Use matching base names, for example: NLPDE_Ch2.pdf and NLPDE_Ch2.ipynb"
  exit 1
fi

[Copilot]

Pull request overview

Adds a PR-triggered GitHub Actions workflow and supporting documentation/instructions to enforce strict chapter PDF ↔ notebook pairing in NLPDE/ and to structure a multi-stage (coverage → interactivity → author loop) review cycle before human review.

Changes:

• Introduces a pull_request workflow that detects changed chapter assets and enforces a 1:1 PDF↔notebook base-name pairing gate.
• Adds reusable agent instruction “contracts” for content coverage, interactivity, and author revision-loop guidance.
• Documents the end-to-end PR workflow and recommends shared preprocessing for consistent multi-agent context.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
.github/workflows/chapter-pr-agent-cycle.yml	Adds PR workflow orchestration, pairing gate, and staged (placeholder) agent jobs.
.github/agent-instructions/content-coverage-agent.md	Defines coverage agent mission/rules and PR-comment output format.
.github/agent-instructions/interactivity-agent.md	Defines interactivity agent mission/rules and PR-comment output format.
.github/agent-instructions/chapter-author-agent.md	Defines author agent workflow and completion gate criteria.
docs/agent-workflow.md	Documents the workflow behavior and recommended shared preprocessing pattern.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.