memory-index-lint

# Memory Index Lint ## Purpose / When to Activate Validate that `index.md` and `archive/superseded-decisions.archive.md` satisfy the discoverability invariants required for AI agents to reliably navigate the memory corpus without missing context. **Activate when:** - Pre-commit (via git hook) - Post-merge (via CI) - Before invoking `memory-archive-superseded` (precondition check) - After invoking `memory-archive-superseded` (post-condition check) - After manual edits to `index.md` or archive files - As part of `memory-index-sync` post-step --- ## Two execution paths ### L4 — Capability-dependent script (preferred when shell available) When the agent has Bash/shell access (Claude Code, Cursor with terminal, daemon `claude -p`) : ```bash python3 scripts/validate-memory-index.py # Exit 0 = CLEAN # Exit 1 = VIOLATIONS (printed to stderr/stdout) # Exit 2 = USAGE / IO error # Strict mode: treat soft warnings as failures (use post-migration) python3 scripts/validate-memory-index.py --strict # Quiet mode: print only final summary line python3 scripts/validate-memory-index.py --quiet ``` Pre-commit hook invokes the script automatically — no manual invocation needed in most flows. ### L3 — Tool-agnostic in-session prompt (fallback when no shell) When the agent is on a surface without shell access, or when invoking the script is not appropriate, perform these checks manually by reading the files : #### Invariant checks (mandatory) 1. **Table column consistency.** For ever...