deep-readlisted

# Deep Read — Codebase Reading Engine Systematic source-code-first analysis protocol. Reads implementations, not just interfaces. Every finding cites `file:line`. **Core principle:** Source code is the source of truth. Documentation lies, comments rot, function names mislead. Read the actual code. A code graph or index (LSP, a code-graph tool) speeds up *locating* code but is itself a possibly-stale snapshot — never ground truth. Verify every graph- or LSP-derived hit against live source before citing it. ## Protocol Process every `/deep-read` invocation through these 6 phases in strict order. Never skip a phase. Gate each phase: do not advance until the gate condition is met. --- ### Phase 1: SCOPE — Define the Reading Target Narrow the target to a tractable area before reading anything. 1. Parse `$ARGUMENTS` as the reading target (module, flow, question, or path) 2. Read CLAUDE.md and MEMORY.md for project context — but treat these as **hints, not truth** 3. Run initial discovery to estimate scope: - `Glob` for relevant file patterns (`**/*.ts`, `**/*.py`, etc.) - `Grep` for key terms from the target description - Count matching files 4. If scope exceeds 50 source files, narrow **deterministically** — this skill runs forked and cannot ask the user (see Fork note): - Rank candidates by centrality (import fan-in within the candidate set), tie-broken by recency (`git log -1 --format=%ct -- <file>`) - Keep the top 49 and drop the rest (the Phase-1 gate be