@saluzi/codegraph 0.1.0 → 0.2.0

package/package.json

@@ -1,24 +1,56 @@
 {
   "name": "@saluzi/codegraph",
-  "version": "0.1.0",
-  "description": "Local-first code intelligence — builds a semantic knowledge graph from any codebase using tree-sitter + SQLite",
-  "main": "./dist/index.js",
-  "types": "./dist/index.d.ts",
+  "version": "0.2.0",
+  "description": "Supercharge AI coding agents with semantic code intelligence — surgical context, fewer tool calls, faster answers. 100% local.",
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "bin": {
+    "codegraph": "./dist/bin/codegraph.js"
+  },
   "files": [
-    "dist/"
+    "dist",
+    "scripts",
+    "README.md"
   ],
+  "scripts": {
+    "build": "tsc && npm run copy-assets && node -e \"require('fs').chmodSync('dist/bin/codegraph.js', 0o755)\"",
+    "preuninstall": "node dist/bin/uninstall.js",
+    "copy-assets": "node -e \"const fs=require('fs');fs.mkdirSync('dist/db',{recursive:true});fs.copyFileSync('src/db/schema.sql','dist/db/schema.sql');fs.mkdirSync('dist/extraction/wasm',{recursive:true});fs.readdirSync('src/extraction/wasm').filter(f=>f.endsWith('.wasm')).forEach(f=>fs.copyFileSync('src/extraction/wasm/'+f,'dist/extraction/wasm/'+f))\"",
+    "dev": "tsc --watch",
+    "cli": "npm run build && node dist/bin/codegraph.js",
+    "test": "vitest run",
+    "test:watch": "vitest",
+    "test:eval": "vitest run __tests__/evaluation/",
+    "eval": "npm run build && npx tsx __tests__/evaluation/runner.ts",
+    "clean": "node -e \"const fs=require('fs');fs.rmSync('dist',{recursive:true,force:true})\""
+  },
   "keywords": [
     "code-intelligence",
-    "tree-sitter",
-    "sqlite",
     "knowledge-graph",
-    "code-analysis"
+    "static-analysis"
   ],
+  "author": "",
   "license": "MIT",
   "dependencies": {
-    "web-tree-sitter": "^0.25.3",
+    "@clack/prompts": "^1.3.0",
+    "commander": "^14.0.2",
+    "fast-string-width": "^3.0.2",
+    "fast-wrap-ansi": "^0.2.0",
     "ignore": "^7.0.5",
+    "jsonc-parser": "^3.3.1",
     "picomatch": "^4.0.3",
-    "jsonc-parser": "^3.3.1"
+    "sisteransi": "^1.0.5",
+    "tree-sitter-wasms": "^0.1.11",
+    "web-tree-sitter": "^0.25.3"
+  },
+  "devDependencies": {
+    "@types/better-sqlite3": "^7.6.0",
+    "@types/node": "^20.19.30",
+    "@types/picomatch": "^4.0.2",
+    "typescript": "^5.0.0",
+    "vitest": "^2.1.9"
+  },
+  "engines": {
+    "node": ">=20.0.0 <25.0.0"
   }
 }

package/scripts/add-lang/bench.sh

@@ -0,0 +1,60 @@
+#!/usr/bin/env bash
+# Add-lang benchmark for ONE repo:
+#   clone -> wipe+index (with the codegraph on PATH) -> verify extraction ->
+#   with/without retrieval A/B (reuses scripts/agent-eval/run-all.sh).
+#
+# Assumes the codegraph dev build is already built + linked on PATH — the skill
+# runs `npm run build && ./scripts/local-install.sh` ONCE before looping repos.
+# The A/B is skipped if extraction fails its critical checks (don't burn $ on a
+# broken extractor); set FORCE_AB=1 to run it anyway.
+#
+# Usage: bench.sh <lang> <repo-name> <repo-url> "<question>" [headless|tmux|all]
+# Env:   CORPUS   corpus dir (default /tmp/codegraph-corpus, shared with agent-eval)
+set -uo pipefail
+
+LANG_TOKEN="${1:?usage: bench.sh <lang> <repo-name> <repo-url> \"<question>\" [mode]}"
+NAME="${2:?repo-name required}"
+URL="${3:?repo-url required}"
+Q="${4:?question required}"
+MODE="${5:-headless}"
+
+HARNESS="$(cd "$(dirname "$0")" && pwd)"
+AGENT_EVAL="$(cd "$HARNESS/../agent-eval" && pwd)"
+CORPUS="${CORPUS:-/tmp/codegraph-corpus}"
+REPO="$CORPUS/$NAME"
+
+command -v codegraph >/dev/null || { echo "no codegraph on PATH (build + ./scripts/local-install.sh first)"; exit 1; }
+
+echo "==================== add-lang bench: $NAME ($LANG_TOKEN) ===================="
+echo "codegraph: $(command -v codegraph) -> $(codegraph --version 2>/dev/null || echo '?')"
+
+# 1. Ensure the repo (shallow clone, reuse if present).
+mkdir -p "$CORPUS"
+if [ -d "$REPO/.git" ]; then
+  echo "→ reusing checkout: $REPO"
+else
+  echo "→ cloning $URL"
+  git clone --depth 1 "$URL" "$REPO" || { echo "git clone failed"; exit 1; }
+fi
+
+# 2. Wipe + index with the binary under test.
+echo "→ wiping .codegraph and indexing"
+rm -rf "$REPO/.codegraph"
+( cd "$REPO" && codegraph init -i ) || { echo "indexing failed"; exit 1; }
+
+# 3. Verify extraction (cheap guard before the paid A/B).
+echo "→ verifying extraction"
+node "$HARNESS/verify-extraction.mjs" "$REPO" "$LANG_TOKEN"
+VERIFY=$?
+
+# 4. Retrieval A/B (skipped if extraction is broken, unless FORCE_AB=1).
+if [ "$VERIFY" -ne 0 ] && [ "${FORCE_AB:-0}" != "1" ]; then
+  echo "→ SKIPPING A/B — extraction failed critical checks (set FORCE_AB=1 to override)"
+else
+  echo "→ retrieval A/B (mode=$MODE)"
+  bash "$AGENT_EVAL/run-all.sh" "$REPO" "$Q" "$MODE"
+fi
+
+echo "==================== bench complete: $NAME (verify exit=$VERIFY) ===================="
+# Exit reflects extraction: 0 = pass/warn, 1 = critical fail, 2 = couldn't read status.
+exit "$VERIFY"

package/scripts/add-lang/check-grammar.mjs

@@ -0,0 +1,75 @@
+#!/usr/bin/env node
+// Verify a tree-sitter grammar wasm is HEALTHY under the project's web-tree-sitter
+// runtime BEFORE writing an extractor. Prints the ABI version and parses a valid
+// sample many times in a multi-grammar context, to catch heap-corruption bugs
+// that silently drop nodes on every parse after the first.
+//
+// Why this exists: the tree-sitter-wasms Lua grammar is ABI 13 and corrupts the
+// shared WASM heap under web-tree-sitter 0.25 — Lua extraction degraded on every
+// file after the first (nested calls/imports vanished). The fix was to vendor the
+// upstream ABI-15 wasm. Run this on any new grammar first; if it FAILs, vendor a
+// newer build instead of using the tree-sitter-wasms one.
+//
+// Usage: node scripts/add-lang/check-grammar.mjs <lang|wasm-path> <valid-sample> [iterations]
+// Exit: 0 healthy, 1 corruption / parse errors, 2 could not run.
+// NOTE: the sample must be SYNTACTICALLY VALID — a broken sample fails for the
+//       wrong reason.
+
+import { readFileSync, existsSync } from 'node:fs';
+import { createRequire } from 'node:module';
+import { Parser, Language } from 'web-tree-sitter';
+
+const require = createRequire(import.meta.url);
+const fail = (code, msg) => { console.error(`[check-grammar] ${msg}`); process.exit(code); };
+
+const [token, sample, iterArg] = process.argv.slice(2);
+if (!token || !sample) fail(2, 'usage: check-grammar.mjs <lang|wasm-path> <valid-sample> [iterations]');
+if (!existsSync(sample)) fail(2, `sample not found: ${sample}`);
+const iters = iterArg ? parseInt(iterArg, 10) : 20;
+
+const SPECIAL = { csharp: 'c_sharp', 'c#': 'c_sharp' };
+function resolveWasm(t) {
+  if (t.endsWith('.wasm')) return existsSync(t) ? t : fail(2, `wasm not found: ${t}`);
+  const base = SPECIAL[t.toLowerCase()] ?? t.toLowerCase();
+  try { return require.resolve(`tree-sitter-wasms/out/tree-sitter-${base}.wasm`); } catch { /* try vendored */ }
+  const vendored = `src/extraction/wasm/tree-sitter-${base}.wasm`;
+  if (existsSync(vendored)) return vendored;
+  return fail(2, `no grammar for "${t}" — not in tree-sitter-wasms and not vendored`);
+}
+
+const wasmPath = resolveWasm(token);
+const source = readFileSync(sample, 'utf8');
+
+try { await Parser.init(); }
+catch { await Parser.init({ locateFile: () => require.resolve('web-tree-sitter/tree-sitter.wasm') }); }
+
+// Load a second, known-good grammar — the corruption surfaces under the
+// multi-grammar runtime that real indexing uses, not a single grammar in isolation.
+try { await Language.load(require.resolve('tree-sitter-wasms/out/tree-sitter-python.wasm')); } catch { /* ok */ }
+
+let language;
+try { language = await Language.load(wasmPath); }
+catch (e) { fail(2, `failed to load ${wasmPath}: ${e.message}`); }
+
+const parser = new Parser();
+parser.setLanguage(language);
+
+let ok = 0, err = 0;
+for (let i = 0; i < iters; i++) {
+  const tree = parser.parse(source);
+  if (tree.rootNode.hasError) err++; else ok++;
+}
+
+console.log(`grammar: ${wasmPath.split('/').pop()}`);
+console.log(`  ABI version: ${language.abiVersion}`);
+console.log(`  parses: ${ok} clean / ${err} with errors (of ${iters})`);
+if (err > 0) {
+  console.log(
+    `RESULT: FAIL — ${err}/${iters} parses produced ERROR trees on a valid sample. ` +
+    `This grammar corrupts under web-tree-sitter; vendor a newer (ABI 14/15) wasm ` +
+    `(see SKILL.md "Find a grammar"). Confirm your sample is syntactically valid first.`
+  );
+  process.exit(1);
+}
+console.log('RESULT: PASS — grammar parses cleanly and reuses safely.');
+process.exit(0);

package/scripts/add-lang/dump-ast.mjs

@@ -0,0 +1,103 @@
+#!/usr/bin/env node
+// Dump the tree-sitter AST for a sample file so you can write a LanguageExtractor
+// mapping. Loads a grammar .wasm directly via web-tree-sitter (the same runtime
+// codegraph uses) — you do NOT need to register the language first.
+//
+// Usage:
+//   node scripts/add-lang/dump-ast.mjs <lang|wasm-path> <sample-file> [--depth=N] [--full]
+// Examples:
+//   node scripts/add-lang/dump-ast.mjs lua sample.lua
+//   node scripts/add-lang/dump-ast.mjs src/extraction/wasm/tree-sitter-zig.wasm a.zig --depth=4
+//
+// Output: an indented AST (named nodes, with field names) followed by a
+// node-type FREQUENCY table. The frequency table is the payoff — it tells you
+// which node types to map to functionTypes / classTypes / importTypes / etc.
+
+import { readFileSync, existsSync } from 'node:fs';
+import { createRequire } from 'node:module';
+import { Parser, Language } from 'web-tree-sitter';
+
+const require = createRequire(import.meta.url);
+const fail = (msg) => { console.error(`[dump-ast] ${msg}`); process.exit(1); };
+
+const argv = process.argv.slice(2);
+const positional = argv.filter((a) => !a.startsWith('--'));
+const [langOrWasm, sampleFile] = positional;
+const depthFlag = argv.find((a) => a.startsWith('--depth='));
+const showAll = argv.includes('--full'); // also print anonymous (token) nodes
+const maxDepth = depthFlag ? parseInt(depthFlag.split('=')[1], 10) : (showAll ? Infinity : 8);
+
+if (!langOrWasm || !sampleFile) {
+  fail('usage: dump-ast.mjs <lang|wasm-path> <sample-file> [--depth=N] [--full]');
+}
+if (!existsSync(sampleFile)) fail(`sample file not found: ${sampleFile}`);
+
+// Language tokens whose tree-sitter-wasms filename differs from the token.
+const WASM_SPECIAL = { csharp: 'c_sharp', 'c#': 'c_sharp' };
+
+function resolveWasm(token) {
+  if (token.endsWith('.wasm')) {
+    if (!existsSync(token)) fail(`wasm not found: ${token}`);
+    return token;
+  }
+  const base = WASM_SPECIAL[token.toLowerCase()] ?? token.toLowerCase();
+  try {
+    return require.resolve(`tree-sitter-wasms/out/tree-sitter-${base}.wasm`);
+  } catch {
+    /* not in tree-sitter-wasms — try a vendored copy */
+  }
+  const vendored = `src/extraction/wasm/tree-sitter-${base}.wasm`;
+  if (existsSync(vendored)) return vendored;
+  fail(
+    `no grammar for "${token}" — not in tree-sitter-wasms and not vendored at ` +
+      `${vendored}. Pass an explicit .wasm path, or vendor one (see SKILL.md "Find a grammar").`
+  );
+}
+
+const wasmPath = resolveWasm(langOrWasm);
+const source = readFileSync(sampleFile, 'utf8');
+
+try {
+  await Parser.init();
+} catch {
+  await Parser.init({ locateFile: () => require.resolve('web-tree-sitter/tree-sitter.wasm') });
+}
+
+let language;
+try {
+  language = await Language.load(wasmPath);
+} catch (e) {
+  fail(`failed to load grammar ${wasmPath}: ${e.message}`);
+}
+
+const parser = new Parser();
+parser.setLanguage(language);
+const tree = parser.parse(source);
+
+const freq = new Map();
+const snippet = (node) => {
+  const t = node.text.replace(/\s+/g, ' ').trim();
+  return t.length > 48 ? `${t.slice(0, 48)}…` : t;
+};
+
+function walk(node, depth, fieldName) {
+  if (node.isNamed) freq.set(node.type, (freq.get(node.type) || 0) + 1);
+  if ((node.isNamed || showAll) && depth <= maxDepth) {
+    const field = fieldName ? `${fieldName}: ` : '';
+    const leaf = node.childCount === 0 ? `  "${snippet(node)}"` : '';
+    console.log(`${'  '.repeat(depth)}${field}${node.type}  @${node.startPosition.row + 1}:${node.startPosition.column}${leaf}`);
+  }
+  for (let i = 0; i < node.childCount; i++) {
+    const child = node.child(i);
+    if (child) walk(child, depth + 1, node.fieldNameForChild(i));
+  }
+}
+
+console.log(`\n# AST for ${sampleFile}  (grammar: ${wasmPath.split('/').pop()})\n`);
+walk(tree.rootNode, 0, null);
+
+console.log('\n# Node-type frequency (named nodes) — map the relevant ones in your extractor:\n');
+[...freq.entries()]
+  .sort((a, b) => b[1] - a[1])
+  .forEach(([type, n]) => console.log(`  ${String(n).padStart(5)}  ${type}`));
+console.log();

package/scripts/add-lang/verify-extraction.mjs

@@ -0,0 +1,70 @@
+#!/usr/bin/env node
+// Sanity-check that codegraph extracted REAL symbols (not just file/import nodes)
+// from a repo for a given language. Exits non-zero on a critical failure so it
+// can drive a write-extractor -> build -> re-check loop.
+//
+// Usage: node scripts/add-lang/verify-extraction.mjs <repo-path> <lang>
+// Reads `codegraph status <repo> --json` using whatever codegraph is on PATH,
+// so it reflects the binary that built the index.
+//
+// Exit codes: 0 = pass or soft-warn, 1 = critical fail, 2 = could not run.
+
+import { execFileSync } from 'node:child_process';
+
+const [repo, lang] = process.argv.slice(2);
+if (!repo || !lang) {
+  console.error('usage: verify-extraction.mjs <repo-path> <lang>');
+  process.exit(2);
+}
+
+let status;
+try {
+  const out = execFileSync('codegraph', ['status', repo, '--json'], { encoding: 'utf8' });
+  status = JSON.parse(out);
+} catch (e) {
+  console.error(`[verify] could not read codegraph status for ${repo}: ${e.message}`);
+  process.exit(2);
+}
+
+// Kinds that prove the extractor mapped AST node types (everything except
+// 'file' and 'import', which codegraph creates structurally for any language).
+const SYMBOL_KINDS = new Set([
+  'module', 'class', 'struct', 'interface', 'trait', 'protocol', 'function',
+  'method', 'property', 'field', 'variable', 'constant', 'enum', 'enum_member',
+  'type_alias', 'namespace', 'route', 'component',
+]);
+
+const byKind = status.nodesByKind || {};
+const langs = status.languages || [];
+const files = status.fileCount || 0;
+const edges = status.edgeCount || 0;
+const symbolKinds = Object.keys(byKind).filter((k) => SYMBOL_KINDS.has(k));
+const symbolCount = symbolKinds.reduce((s, k) => s + byKind[k], 0);
+
+const checks = [];
+const add = (severity, ok, label, detail) => checks.push({ severity, ok, label, detail });
+
+add('critical', status.initialized === true, 'index initialized', `initialized=${status.initialized}`);
+add('critical', langs.includes(lang), `language "${lang}" detected`, `languages=[${langs.join(', ')}]`);
+add('critical', symbolCount > 0, 'structural symbols extracted', `${symbolCount} symbols (${symbolKinds.join(', ') || 'NONE — only file/import nodes!'})`);
+add('soft', symbolCount >= files, 'symbol density >= 1/file', `${symbolCount} symbols across ${files} files`);
+add('soft', edges > files, 'edges resolved', `${edges} edges across ${files} files`);
+
+console.log(`\n# Extraction check — ${repo}  (lang=${lang}, backend=${status.backend})`);
+console.log(`  files=${files} nodes=${status.nodeCount} edges=${edges}`);
+console.log(`  nodesByKind: ${JSON.stringify(byKind)}\n`);
+for (const c of checks) console.log(`  ${c.ok ? '✓' : '✗'} ${c.label} — ${c.detail}`);
+
+const critical = checks.filter((c) => !c.ok && c.severity === 'critical');
+const soft = checks.filter((c) => !c.ok && c.severity === 'soft');
+console.log();
+if (critical.length) {
+  console.log(`RESULT: FAIL (${critical.length} critical) — extractor or grammar wiring is broken. Re-run dump-ast.mjs and fix the node-type mappings.`);
+  process.exit(1);
+}
+if (soft.length) {
+  console.log(`RESULT: WARN (${soft.length} soft) — extraction works but looks thin; inspect the counts above.`);
+  process.exit(0);
+}
+console.log('RESULT: PASS — extraction looks healthy.');
+process.exit(0);

package/scripts/agent-eval/ab-adoption.sh

@@ -0,0 +1,91 @@
+#!/usr/bin/env bash
+# Does the agent PICK codegraph_node to read a file, vs the built-in Read tool?
+# Build A/B: NEW build (HEAD, codegraph_node has Read parity) vs BASELINE build
+# (a ref where it doesn't), BOTH codegraph-attached + pre-warmed, same task. The
+# metric is tool CHOICE: Read calls vs codegraph_node[file] calls per run.
+#
+# Usage: ab-adoption.sh <indexed-repo> "<task>" [runs-per-arm] [baseline-ref]
+# Env: AGENT_EVAL_OUT (default: /tmp/ab-adoption)
+set -uo pipefail
+TARGET="${1:?usage: ab-adoption.sh <indexed-repo> \"<task>\" [runs] [baseline-ref]}"
+TASK="${2:?task required}"
+RUNS="${3:-2}"
+BASE_REF="${4:-HEAD~1}"
+ENGINE="$(cd "$(dirname "$0")/../.." && pwd)"
+BIN="$ENGINE/dist/bin/codegraph.js"
+OUT="${AGENT_EVAL_OUT:-/tmp/ab-adoption}"
+
+command -v claude >/dev/null || { echo "claude CLI not on PATH"; exit 1; }
+[ -d "$TARGET/.codegraph" ] || { echo "target not indexed: run 'codegraph init $TARGET' first"; exit 1; }
+git -C "$ENGINE" diff --quiet && git -C "$ENGINE" diff --cached --quiet || { echo "engine has uncommitted changes — commit/stash first"; exit 1; }
+CHANGED=$(git -C "$ENGINE" diff --name-only "$BASE_REF" HEAD -- src 2>/dev/null)
+[ -n "$CHANGED" ] || { echo "no src/ changes between $BASE_REF and HEAD"; exit 1; }
+
+cleanup() {
+  pkill -9 -f "serve --mcp --path $OUT/" 2>/dev/null
+  git -C "$ENGINE" checkout HEAD -- $CHANGED 2>/dev/null
+  ( cd "$ENGINE" && npm run build >/dev/null 2>&1 )
+}
+trap cleanup EXIT
+mkdir -p "$OUT"
+echo "###### target=$TARGET  runs/arm=$RUNS  baseline=$BASE_REF"
+echo "###### changed: $(echo "$CHANGED" | tr '\n' ' ')"
+echo "###### task=$TASK"; echo
+
+prewarm() {
+  pkill -9 -f "serve --mcp --path $1" 2>/dev/null
+  CODEGRAPH_DAEMON_IDLE_TIMEOUT_MS=1800000 node "$BIN" serve --mcp --path "$1" </dev/null >/dev/null 2>&1 &
+  node -e 'const fs=require("fs");let n=0;const t=setInterval(()=>{if(fs.existsSync(process.argv[1]+"/.codegraph/daemon.sock")){clearInterval(t);process.exit(0)}if(n++>150){clearInterval(t);process.exit(1)}},100)' "$1" >/dev/null 2>&1
+}
+
+# Per-run tool-choice counts: Read vs codegraph_node[file] vs [symbol].
+count() {
+  node -e '
+    const fs=require("fs");
+    const lines=fs.readFileSync(process.argv[1],"utf8").split("\n").filter(Boolean);
+    let read=0,cgFile=0,cgSym=0,cgOther=0,exposed="?";
+    for(const l of lines){try{const o=JSON.parse(l);
+      if(o.type==="system"&&o.subtype==="init"){exposed=(o.tools||[]).filter(t=>/codegraph/.test(t)).length;}
+      const blocks=o.message?.content||[];
+      for(const b of (Array.isArray(blocks)?blocks:[])){
+        if(b.type!=="tool_use")continue;
+        if(b.name==="Read")read++;
+        else if(b.name==="mcp__codegraph__codegraph_node"){ if(b.input&&b.input.symbol)cgSym++; else cgFile++; }
+        else if(/mcp__codegraph__/.test(b.name))cgOther++;
+      }
+    }catch{}}
+    console.log(`    Read=${read}  codegraph_node[file]=${cgFile}  codegraph_node[symbol]=${cgSym}  other_cg=${cgOther}  (cg exposed=${exposed})`);
+  ' "$1"
+}
+
+run_arm() { # label, N
+  local label="$1" n="$2"
+  local c="$OUT/mcp-$label.json"
+  for i in $(seq 1 "$n"); do
+    local tgt="$OUT/t-$label-$i"
+    rm -rf "$tgt"
+    rsync -a --exclude node_modules --exclude .git --exclude dist --exclude .codegraph "$TARGET/" "$tgt/"
+    node "$BIN" init "$tgt" >/dev/null 2>&1
+    printf '{"mcpServers":{"codegraph":{"command":"env","args":["CODEGRAPH_WASM_RELAUNCHED=1","node","%s","serve","--mcp","--path","%s"]}}}' "$BIN" "$tgt" > "$c"
+    prewarm "$tgt"
+    echo "----- [$label] run $i -----"
+    ( cd "$tgt" && claude -p "$TASK" \
+        --output-format stream-json --verbose --permission-mode bypassPermissions \
+        --model "${MODEL:-sonnet}" --effort "${EFFORT:-high}" --max-budget-usd 4 --strict-mcp-config --mcp-config "$c" \
+        </dev/null > "$OUT/run-$label-$i.jsonl" 2>"$OUT/run-$label-$i.err" )
+    count "$OUT/run-$label-$i.jsonl"
+    pkill -9 -f "serve --mcp --path $tgt" 2>/dev/null
+  done
+  echo
+}
+
+echo "== NEW build (HEAD: codegraph_node has Read parity) =="
+( cd "$ENGINE" && npm run build >/dev/null 2>&1 ) && echo "built"
+run_arm new "$RUNS"
+
+echo "== BASELINE build ($BASE_REF) =="
+git -C "$ENGINE" checkout "$BASE_REF" -- $CHANGED
+( cd "$ENGINE" && npm run build >/dev/null 2>&1 ) && echo "built"
+run_arm baseline "$RUNS"
+
+echo "###### DONE — compare [new] vs [baseline]: does codegraph_node[file] rise / Read fall? Logs: $OUT"

package/scripts/agent-eval/ab-hook.sh

@@ -0,0 +1,86 @@
+#!/usr/bin/env bash
+# A/B the PreToolUse(Read) REDIRECT hook (P1): does steering Read → codegraph_node
+# file-view actually move the agent off Read during implementation? BOTH arms use
+# the CURRENT build with codegraph attached and pre-warmed; the only difference is
+# the hook. Isolates the hook's behavioral effect from the build/file-view change
+# (use ab-new-vs-baseline.sh for the build A/B).
+#
+#   arm [nohook] — codegraph on, no hook   (does the better file-view get picked on its own?)
+#   arm [hook]   — codegraph on, + redirect hook   (does routing close it?)
+#
+# Reliable attach (works nested): each arm pre-warms a persistent daemon and skips
+# the startup re-exec (CODEGRAPH_WASM_RELAUNCHED=1), so claude connects before the
+# agent's first turn. Judge by ACTUAL codegraph usage in parse-run.mjs's "by type",
+# not claude's init snapshot (which can read pending even when it then connects).
+#
+# Usage: ab-hook.sh <indexed-repo> "<implementation task>" [runs-per-arm]
+#   <indexed-repo>  a repo with a .codegraph index (copied per arm; never mutated)
+#   "<task>"        a GENUINELY-NEW implementation task (verify it isn't already done)
+#   [runs-per-arm]  default 2 (n=1 is noisy — the doctrine says >=2)
+# Env: AGENT_EVAL_OUT (default: /tmp/ab-hook)
+set -uo pipefail
+
+TARGET="${1:?usage: ab-hook.sh <indexed-repo> \"<task>\" [runs-per-arm]}"
+TASK="${2:?task required}"
+RUNS="${3:-2}"
+ENGINE="$(cd "$(dirname "$0")/../.." && pwd)"
+BIN="$ENGINE/dist/bin/codegraph.js"
+HOOK="$ENGINE/scripts/agent-eval/redirect-read-hook.sh"
+OUT="${AGENT_EVAL_OUT:-/tmp/ab-hook}"
+PARSE="$ENGINE/scripts/agent-eval/parse-run.mjs"
+
+command -v claude >/dev/null || { echo "claude CLI not on PATH"; exit 1; }
+command -v jq >/dev/null || { echo "jq not on PATH (the hook needs it)"; exit 1; }
+[ -d "$TARGET/.codegraph" ] || { echo "target not indexed: run 'codegraph init $TARGET' first"; exit 1; }
+chmod +x "$HOOK"
+
+cleanup() { pkill -9 -f "serve --mcp --path $OUT/" 2>/dev/null; }
+trap cleanup EXIT
+
+mkdir -p "$OUT"
+echo "###### engine=$ENGINE"
+echo "###### target=$TARGET   runs/arm=$RUNS"
+echo "###### task=$TASK"
+echo
+
+( cd "$ENGINE" && npm run build >/dev/null 2>&1 ) && echo "built"
+
+# A settings file carrying ONLY the PreToolUse(Read) redirect hook.
+HOOK_SETTINGS="$OUT/hook-settings.json"
+jq -n --arg cmd "bash $HOOK" \
+  '{hooks:{PreToolUse:[{matcher:"Read",hooks:[{type:"command",command:$cmd}]}]}}' > "$HOOK_SETTINGS"
+
+prewarm() { # target — spawn a persistent daemon and wait for its socket
+  pkill -9 -f "serve --mcp --path $1" 2>/dev/null
+  CODEGRAPH_DAEMON_IDLE_TIMEOUT_MS=1800000 node "$BIN" serve --mcp --path "$1" </dev/null >/dev/null 2>&1 &
+  node -e 'const fs=require("fs");let n=0;const t=setInterval(()=>{if(fs.existsSync(process.argv[1]+"/.codegraph/daemon.sock")){clearInterval(t);process.exit(0)}if(n++>150){clearInterval(t);process.exit(1)}},100)' "$1" \
+    && echo "  daemon warm: $1" || echo "  WARN: daemon never bound for $1"
+}
+
+run_one() { # arm-label, run-index, use-hook(0|1)
+  local label="$1" idx="$2" hook="$3"
+  local tgt="$OUT/t-$label-$idx" c="$OUT/mcp-$label.json"
+  rm -rf "$tgt"
+  rsync -a --exclude node_modules --exclude .git --exclude dist --exclude .codegraph "$TARGET/" "$tgt/"
+  node "$BIN" init "$tgt" >/dev/null 2>&1
+  printf '{"mcpServers":{"codegraph":{"command":"env","args":["CODEGRAPH_WASM_RELAUNCHED=1","node","%s","serve","--mcp","--path","%s"]}}}' "$BIN" "$tgt" > "$c"
+  prewarm "$tgt"
+  local extra=()
+  [ "$hook" = "1" ] && extra=(--settings "$HOOK_SETTINGS")
+  echo "----- [$label] run $idx -----"
+  # ${extra[@]+...} guard: bash 3.2 (macOS) under `set -u` errors on an empty
+  # array expansion otherwise, which would skip the no-hook arm's claude run.
+  ( cd "$tgt" && claude -p "$TASK" \
+      --output-format stream-json --verbose --permission-mode bypassPermissions \
+      --model "${MODEL:-sonnet}" --effort "${EFFORT:-high}" --max-budget-usd 4 --strict-mcp-config --mcp-config "$c" ${extra[@]+"${extra[@]}"} \
+      </dev/null > "$OUT/run-$label-$idx.jsonl" 2>"$OUT/run-$label-$idx.err" )
+  node "$PARSE" "$OUT/run-$label-$idx.jsonl" 2>&1 | grep -E "by type|Result" || echo "  (parse failed — see $OUT/run-$label-$idx.jsonl)"
+  pkill -9 -f "serve --mcp --path $tgt" 2>/dev/null
+  echo
+}
+
+for i in $(seq 1 "$RUNS"); do run_one nohook "$i" 0; done
+for i in $(seq 1 "$RUNS"); do run_one hook   "$i" 1; done
+
+echo "###### DONE. Compare [nohook] vs [hook] 'by type' — Read should fall and"
+echo "###### mcp__codegraph__codegraph_node should rise in the [hook] arm. Logs: $OUT"

package/scripts/agent-eval/ab-impl.sh

@@ -0,0 +1,78 @@
+#!/usr/bin/env bash
+# Sufficiency A/B for an IMPLEMENTATION task (the agent edits): when it uses
+# codegraph (explore/node) to understand before editing, does it still Read? Like
+# ab-sufficiency.sh but copies+indexes a FRESH target per run (the agent mutates
+# it), so runs don't see each other's edits.
+#
+# WITH codegraph (pre-warmed) vs WITHOUT (empty MCP), N runs each. Reports
+# explore/node vs Read/Grep + the files Read, and whether the build still passes.
+#
+# Usage: ab-impl.sh <indexed-repo> "<task>" [runs] [build-cmd]
+# Env: AGENT_EVAL_OUT (default: /tmp/ab-impl)
+set -uo pipefail
+REPO="${1:?usage: ab-impl.sh <indexed-repo> \"<task>\" [runs] [build-cmd]}"
+Q="${2:?task required}"
+RUNS="${3:-2}"
+BUILD_CMD="${4:-}"
+ENGINE="$(cd "$(dirname "$0")/../.." && pwd)"
+BIN="$ENGINE/dist/bin/codegraph.js"
+OUT="${AGENT_EVAL_OUT:-/tmp/ab-impl}"
+command -v claude >/dev/null || { echo "claude CLI not on PATH"; exit 1; }
+[ -d "$REPO/.codegraph" ] || { echo "no .codegraph index at $REPO"; exit 1; }
+cleanup(){ pkill -9 -f "serve --mcp --path $OUT/" 2>/dev/null; }
+trap cleanup EXIT
+mkdir -p "$OUT"
+( cd "$ENGINE" && npm run build >/dev/null 2>&1 ) && echo "built engine"
+echo "###### repo=$REPO  runs/arm=$RUNS"
+echo "###### task=$Q"; echo
+echo '{"mcpServers":{}}' > "$OUT/mcp-empty.json"
+
+prewarm(){
+  pkill -9 -f "serve --mcp --path $1" 2>/dev/null
+  CODEGRAPH_DAEMON_IDLE_TIMEOUT_MS=1800000 node "$BIN" serve --mcp --path "$1" </dev/null >/dev/null 2>&1 &
+  node -e 'const fs=require("fs");let n=0;const t=setInterval(()=>{if(fs.existsSync(process.argv[1]+"/.codegraph/daemon.sock")){clearInterval(t);process.exit(0)}if(n++>150){clearInterval(t);process.exit(1)}},100)' "$1" >/dev/null 2>&1
+}
+
+analyze(){
+  node -e '
+    const fs=require("fs");
+    const L=fs.readFileSync(process.argv[1],"utf8").split("\n").filter(Boolean);
+    let ex=0,nf=0,ns=0,oc=0,gr=0,ed=0,exposed="?";const reads=[];
+    for(const l of L){try{const o=JSON.parse(l);
+      if(o.type==="system"&&o.subtype==="init")exposed=(o.tools||[]).filter(t=>/codegraph/.test(t)).length;
+      for(const b of (o.message?.content||[])){if(b.type!=="tool_use")continue;
+        if(b.name==="mcp__codegraph__codegraph_explore")ex++;
+        else if(b.name==="mcp__codegraph__codegraph_node"){if(b.input&&b.input.symbol)ns++;else nf++;}
+        else if(/mcp__codegraph__/.test(b.name))oc++;
+        else if(b.name==="Read")reads.push((b.input?.file_path||"").split("/").pop());
+        else if(b.name==="Grep")gr++;
+        else if(b.name==="Edit"||b.name==="Write")ed++;
+      }}catch{}}
+    console.log(`    explore=${ex} node[sym]=${ns} node[file]=${nf} other_cg=${oc} | Read=${reads.length}${reads.length?" ("+reads.join(", ")+")":""} Grep=${gr} Edit=${ed}  [cg exposed=${exposed}]`);
+  ' "$1"
+}
+
+run(){ # label, withCodegraph(0/1)
+  local label="$1" wcg="$2"
+  for i in $(seq 1 "$RUNS"); do
+    local tgt="$OUT/t-$label-$i" cfg="$OUT/mcp-$label.json"
+    rm -rf "$tgt"
+    rsync -a --exclude node_modules --exclude .git --exclude dist --exclude .codegraph "$REPO/" "$tgt/"
+    node "$BIN" init "$tgt" >/dev/null 2>&1
+    if [ "$wcg" = "1" ]; then
+      printf '{"mcpServers":{"codegraph":{"command":"env","args":["CODEGRAPH_WASM_RELAUNCHED=1","node","%s","serve","--mcp","--path","%s"]}}}' "$BIN" "$tgt" > "$cfg"
+      prewarm "$tgt"
+    else cp "$OUT/mcp-empty.json" "$cfg"; fi
+    ( cd "$tgt" && claude -p "$Q" --output-format stream-json --verbose \
+        --permission-mode bypassPermissions --model "${MODEL:-sonnet}" --effort "${EFFORT:-high}" --max-budget-usd 4 \
+        --strict-mcp-config --mcp-config "$cfg" </dev/null > "$OUT/$label-$i.jsonl" 2>"$OUT/$label-$i.err" )
+    echo "[$label] run $i:"; analyze "$OUT/$label-$i.jsonl"
+    if [ -n "$BUILD_CMD" ]; then ( cd "$tgt" && eval "$BUILD_CMD" >/dev/null 2>&1 && echo "      build: PASS" || echo "      build: FAIL" ); fi
+    pkill -9 -f "serve --mcp --path $tgt" 2>/dev/null
+  done
+  echo
+}
+
+echo "== WITH codegraph =="; run with 1
+echo "== WITHOUT (Read/Grep only) =="; run without 0
+echo "###### DONE: $OUT"