v0.2.0: Proactive Context Loading

What's New

Proactive context loading - Load context before you edit, not just enforce after.

New Features

• core/print-context.sh - Portable context lookup (Git + Bash only)

  • Dynamic tier detection from CODE_DOC_MAP.md
  • Shows doc-set, required reading, and invariants
  • UNMAPPED state distinct from Tier C

• protocols/SESSION_PROTOCOL.md - Version sync + cognitive loading order

  • Pre-work checklist
  • Context loading priority table

• protocols/AGENT_PROTOCOL.md - Updated with print-context.sh integration

  • Bug verification rule (no reports without code proof)
  • Streamlined compliance requirements

• core/templates/project-context.template.md - Project memory template

• examples/doc-systems/ - Testbed demonstrating tier classification

Usage

./LivingDocFramework/core/print-context.sh src/api/auth.py

Output:

File: src/api/auth.py
Tier: A (Critical)
Doc-Set: docs/api
Map: docs/api/CODE_DOC_MAP.md

Required Reading:
  1. docs/api/INVARIANTS.md
  2. docs/api/CODE_DOC_MAP.md

Invariants:
  - INV-AUTH-001: Retry attempts bounded
  - INV-AUTH-002: Password never logged

Design Principles (v0.2)

1. Tool-agnostic core (Git + Bash only, no vendor lock-in)
2. Dynamic, not hardcoded (tier from CODE_DOC_MAP.md, not heuristics)
3. Derived, not magic (doc mapping from structure)
4. Optional integrations (Claude Code adapter separate, not core)
5. Parseable contracts (machine-readable with defined structure)

---

Full Changelog: v1.2.2...v0.2.0