weavatrix 0.3.8 → 0.3.10

package/README.md

@@ -131,6 +131,18 @@ Every finding is review evidence, never an auto-delete verdict: `find_dead_code`
 `run_audit category=unused` always return `REVIEW_REQUIRED` with `autoDelete:false`. Typecheck, tests
 and runtime checks remain the release authority.
 
+## Always-fresh graph
+
+There is no watcher daemon to run and no manual refresh step: every graph/health call reconciles the
+graph before answering. A Git-token freshness probe (HEAD + dirty/untracked content, debounced 2 s)
+decides whether anything changed; when it did, a bounded incremental refresh reparses only the changed
+files plus their reverse importers (≤ 24 changed / ≤ 80 reparsed JS/TS files) and merges the scoped
+result into the previous graph under a file lock. Config/lockfile edits, export-surface changes,
+barrel files and non-JS/TS languages fall back to a full rebuild — correctness always wins over speed.
+Each refreshed answer carries a structured `refresh` record (`none` / `incremental` / `full`, changed
+file count), so an agent can tell exactly which repository state it is reasoning about. The same
+guarantees hold across concurrent MCP clients sharing one canonical graph.
+
 ## Benchmarks
 
 Two gates ship in the repository:
@@ -178,6 +190,11 @@ JavaScript · TypeScript · TSX · Python · Go · Java · C# · Rust · HTML ·
 [web-tree-sitter](https://github.com/tree-sitter/tree-sitter) WASM grammars; no Python install and no
 native compilation.
 
+Test surfaces are classified per file (path conventions plus `.weavatrix.json` overrides) and, for
+Rust, per symbol: `#[cfg(test)]` modules and `#[test]`/`#[bench]` items inside production `.rs` files
+carry a node-level `test_surface` flag, so dead-code, query, hot-path and hub tools treat them as
+tests rather than production code.
+
 ## Development
 
 ```sh

package/docs/releases/v0.3.10.md

@@ -0,0 +1,76 @@
+# Weavatrix 0.3.10
+
+0.3.10 closes the honesty gaps a production evaluation surfaced: Rust inline
+tests are no longer reported as production code, endpoint tools are
+production-first everywhere, and semantic-precision statuses say exactly how
+much was actually proven. Every fix landed with regression tests and survived
+an adversarial multi-agent review of the full changeset.
+
+## Node-level Rust test classification
+
+Rust puts unit tests inside production files, so path classification alone
+misreported them everywhere. The extractor now proves test-only compilation —
+`#[cfg(test)]` / `#[cfg(all(test, ...))]` on items, impl blocks and modules,
+`#[test]`/`#[tokio::test]`-style harness attributes and `#[bench]` functions,
+inner `#![cfg(test)]` attributes at file and module-body scope, and attributes
+separated from their item by comments — and stamps a `test_surface` flag on the
+node. Deliberately conservative: `#[cfg(any(test, ...))]` also ships in
+production builds and stays unclassified.
+
+Every consumer honors the flag like a path-classified test: `find_dead_code`
+(candidates and test-only consumer resolution), `query_graph` seeds and
+traversal, `hot_path_review`, `god_nodes`, `verified_change` task anchors,
+`no-tests`/`tests-only` build modes, hosted sync payloads and evidence
+snapshots. A test-term question or `include_tests`/`include_classified` opts
+back in. `extractorSchemaV` 5 → 6; cached graphs rebuild once, automatically.
+
+## Production-first endpoint tools
+
+- `list_endpoints` suppresses endpoints declared in classified
+  test/e2e/generated/mock/story/docs/benchmark/temp paths by default, with a
+  count note and `include_classified` opt-in (auto-enabled on tests-only
+  graphs). Opted-in rows carry `[classified:...]` tags.
+- `trace_endpoint` filters classified handler twins before its ambiguity
+  check, ranks remaining same-named candidates by the import binding at the
+  route file (below same-directory proximity — a file-level edge proves
+  coupling, not identity), and reports `importBoundAtRoute`/`pathClasses` per
+  candidate when it still fails closed.
+- New `handler_file` parameter resolves genuinely symmetric candidates: it
+  narrows the full candidate pool before scoring, prefers exact repo-relative
+  matches over path-suffix matches, and compares case-insensitively.
+
+## Honest semantic-precision statuses
+
+`COMPLETE` previously meant only "the bounded request finished", producing
+`COMPLETE (0 EXACT_LSP edges)` on Rust repositories and `COMPLETE; exact
+absence was not proven` in `get_dependents`. Now:
+
+- A new `NONE` state covers the zero-eligible-targets outcome; `UNAVAILABLE`
+  stays reserved for provider failure. `PRECISION_OVERLAY_V` 4 → 5 discards
+  old-vocabulary sidecars.
+- `COMPLETE` requires every selected target actually probed.
+- `open_repo` and the build log print the `graph_stats`-style coverage
+  fraction: `PARTIAL — 32/1373 bounded target(s) queried (truncated)`.
+- `get_dependents` never pairs a state word with a contradicting note, and
+  keeps `NONE`/`UNAVAILABLE` truthful instead of collapsing them into
+  `PARTIAL`.
+
+## Output fidelity
+
+- `trace_api_contract` records per-endpoint method mismatches with sample
+  callers and always shows a mismatch block, so `*_WITH_METHOD_MISMATCHES`
+  verdicts stay backed by visible rows under any `top_n`.
+- Structure findings and hosted evidence snapshots stop reporting idiomatic
+  Rust `mod.rs`/`lib.rs`/`main.rs` parent-child module-tree cycles as
+  compile-time coupling (counted as `rustModuleTreeCouplings`; genuine
+  cross-directory cycles still report).
+- `query_graph` edge lines and `shortest_path` hops widen repeated or generic
+  file basenames (`mod.rs`, `index.*`, `__init__.py`) to their last two path
+  segments.
+
+## Documentation
+
+The always-fresh graph contract — debounced Git freshness probe plus bounded
+incremental refresh on every graph/health call, no watcher process required —
+is now documented in the README and the agent skill, together with the Rust
+test-surface classification rules.

package/docs/releases/v0.3.9.md

@@ -0,0 +1,36 @@
+# Weavatrix 0.3.9
+
+0.3.9 consolidates every subprocess launch behind a small number of audited
+chokepoints. Tool behavior and output are unchanged — verified byte-identical
+against 0.3.8 on real repositories.
+
+## Single-chokepoint subprocess execution
+
+Git and other subprocess launches were scattered across roughly fifteen modules,
+each with its own near-identical spawn wrapper. They now route through two
+focused modules:
+
+- **git** — every git command runs through `src/git-exec.js`: `runGit` for
+  synchronous run-and-collect calls and `boundedGitCommand` for the streaming
+  history collector.
+- **everything else** — ripgrep resolution and search, `tar` archive extraction,
+  and `where`/`which` binary probing run through `src/process.js`: the existing
+  async `runCommand` plus a new synchronous `runCommandSync`.
+
+Every launch now applies the same credential-stripped child environment,
+`windowsHide`, bounded timeout, and `-C <repoRoot>`/`cwd` containment in one
+place. The number of modules importing `node:child_process` drops from fifteen
+to four — the two runners plus the long-lived language-server client — removing
+duplicated wrappers and making the whole process surface auditable from a single
+point.
+
+This is a pure refactor: graph construction, change impact, Health, history and
+search produce identical results. A byte-level comparison of the entire graph
+(every node, edge and external import, hashed with SHA-256) against the published
+0.3.8 matched exactly on the analytics, controller-rest-api and Weavatrix
+repositories.
+
+## Minor
+
+Removed two redundant module imports (`node:perf_hooks`, `node:process`) that
+re-imported values already available as Node globals.

package/package.json

@@ -1,8 +1,8 @@
 {
   "name": "weavatrix",
-  "version": "0.3.8",
+  "version": "0.3.10",
   "type": "module",
-  "description": "Local repository intelligence MCP for AI coding agents: reusable architecture graph for fast application understanding, Health, dead code, clones, history, blast radius and architecture safeguards.",
+  "description": "Local repository intelligence MCP for AI coding agents: an always-fresh architecture graph with production-first evidence — blast radius, dead code, endpoints, clones, history, Health and architecture safeguards, with honestly labeled precision.",
   "author": "Sergii Ziborov <sergii.ziborov@gmail.com>",
   "license": "MIT",
   "homepage": "https://weavatrix.com",
@@ -42,6 +42,8 @@
     "docs/releases/v0.3.6.md",
     "docs/releases/v0.3.7.md",
     "docs/releases/v0.3.8.md",
+    "docs/releases/v0.3.9.md",
+    "docs/releases/v0.3.10.md",
     "docs/releases/v0.2.19.md",
     "README.md",
     "SECURITY.md",

package/scripts/run-agent-task-benchmark.mjs

@@ -1,6 +1,5 @@
 import {readFileSync} from 'node:fs'
 import {pathToFileURL} from 'node:url'
-import {performance} from 'node:perf_hooks'
 import {retrieveTaskContext} from '../src/analysis/task-retrieval.js'
 
 function graphFixture() {

package/skill/SKILL.md

@@ -142,12 +142,15 @@ interface dispatch, reflection, or runtime behavior.
 - **Evidence, not verdicts**: treat audit, hub, orphan and duplicate output as hypotheses. Confirm a
   finding in source and check framework/runtime conventions before deleting, merging or redesigning
   code. A same-name/different-body pair is a divergence candidate, not proof of duplication.
-- **Freshness**: graph/health calls automatically reconcile the active graph, while cross-repository
-  tracing reconciles every selected registered graph. Read the structured `refresh` /
-  `graphReconciliation` status: `none`, `incremental`, `full`, or explicitly `PARTIAL`. Use
-  `rebuild_graph` only when automatic reconciliation reports a fallback/error or when intentionally
-  changing build mode. A normal `open_repo` builds missing graphs and upgrades legacy schemas;
-  `build:false` deliberately refuses that upgrade.
+- **Freshness**: the graph is always fresh at answer time — no watcher process and no manual refresh
+  step exists or is needed. Every graph/health call runs a debounced Git freshness probe and, when the
+  repository changed, refreshes before answering (bounded incremental reparse of changed files plus
+  reverse importers for JS/TS; full rebuild for config/export-surface/barrel changes and other
+  languages). Cross-repository tracing reconciles every selected registered graph the same way. Read
+  the structured `refresh` / `graphReconciliation` status: `none`, `incremental`, `full`, or
+  explicitly `PARTIAL`. Use `rebuild_graph` only when automatic reconciliation reports a
+  fallback/error or when intentionally changing build mode. A normal `open_repo` builds missing graphs
+  and upgrades legacy schemas; `build:false` deliberately refuses that upgrade.
 - **Ambiguity**: `get_node`/`get_neighbors`/`get_dependents` disclose `matched N nodes; using the
   best-connected` — read that note before trusting the answer; pass an exact node id to pin it.
 - **Runtime versus compile time**: keep runtime cycles separate from TypeScript type-only and
@@ -169,6 +172,10 @@ interface dispatch, reflection, or runtime behavior.
   roots. Dead-code/clone/audit review also suppresses `benchmarks/**` and `**/__temp/**` as classified
   non-production surfaces. A verified production benchmark can opt back in narrowly through
   `.weavatrix.json` `classify.product`, for example `{"classify":{"product":["benchmarks/core/**"]}}`.
+  Rust inline tests (`#[cfg(test)]` modules, `#[test]`/`#[bench]` items) are classified per symbol via
+  the node-level `test_surface` flag even though they live in production `.rs` files; dead-code,
+  query, hot-path and hub tools treat them like path-classified tests (`include_tests` /
+  `include_classified` and a test-term question opt back in).
 - **Architectural queries**: broad bootstrap/tool-execution/routing questions rank conventional
   production and graph-declared entry points ahead of classified docs, sites, benchmarks and
   fixtures. Production-first classification also applies during traversal. A class term in the

package/src/analysis/change-classification.js

@@ -1,6 +1,5 @@
 // Symbol-aware git-diff classification for change_impact.
-import {spawnSync} from 'node:child_process'
-import {childProcessEnv} from '../child-env.js'
+import {runGit} from '../git-exec.js'
 import {createPathClassifier} from '../path-classification.js'
 import {parseZeroContextDiff} from './change-classification/diff-parser.js'
 import {
@@ -33,13 +32,10 @@ function recoverBoundedDiff(repoRoot, base, files, limits) {
             fallbackFiles.push(...files.slice(index))
             break
         }
-        const result = spawnSync('git', [
-            '-C', repoRoot, 'diff', '--no-ext-diff', '--find-renames', '--no-color',
+        const result = runGit(repoRoot, [
+            'diff', '--no-ext-diff', '--find-renames', '--no-color',
             '--unified=0', String(base), '--', files[index],
-        ], {
-            encoding: 'utf8', windowsHide: true, timeout: Math.min(2_000, remainingMs),
-            maxBuffer: remainingBytes + 1, env: childProcessEnv(),
-        })
+        ], {timeout: Math.min(2_000, remainingMs), maxBuffer: remainingBytes + 1})
         const chunk = String(result.stdout || '')
         if (result.status !== 0 || !chunk || Buffer.byteLength(chunk) > remainingBytes) {
             fallbackFiles.push(files[index])
@@ -51,11 +47,8 @@ function recoverBoundedDiff(repoRoot, base, files, limits) {
 }
 
 function runGitDiff(repoRoot, base, limits) {
-    const args = ['-C', repoRoot, 'diff', '--no-ext-diff', '--find-renames', '--no-color', '--unified=0', String(base), '--']
-    const result = spawnSync('git', args, {
-        encoding: 'utf8', windowsHide: true, timeout: 12_000,
-        maxBuffer: limits.maxDiffBytes + 1, env: childProcessEnv(),
-    })
+    const args = ['diff', '--no-ext-diff', '--find-renames', '--no-color', '--unified=0', String(base), '--']
+    const result = runGit(repoRoot, args, {timeout: 12_000, maxBuffer: limits.maxDiffBytes + 1})
     const output = String(result.stdout || '')
     const oversized = result.error?.code === 'ENOBUFS' || Buffer.byteLength(output) > limits.maxDiffBytes
     if (result.status === 0 && !oversized) return {available: true, text: output, error: null}
@@ -63,12 +56,9 @@ function runGitDiff(repoRoot, base, limits) {
     let fallbackTruncated = false
     let recoveredText = ''
     if (oversized) {
-        const names = spawnSync('git', [
-            '-C', repoRoot, 'diff', '--name-only', '-z', '--no-ext-diff', '--find-renames', String(base), '--',
-        ], {
-            encoding: 'utf8', windowsHide: true, timeout: 12_000,
-            maxBuffer: Math.max(64 * 1024, limits.maxFiles * 4_096), env: childProcessEnv(),
-        })
+        const names = runGit(repoRoot, [
+            'diff', '--name-only', '-z', '--no-ext-diff', '--find-renames', String(base), '--',
+        ], {timeout: 12_000, maxBuffer: Math.max(64 * 1024, limits.maxFiles * 4_096)})
         if (names.status === 0) {
             const all = String(names.stdout || '').split('\0').map(normalizeChangePath).filter(Boolean)
             const boundedFiles = [...new Set(all)].sort((a, b) => a.localeCompare(b)).slice(0, limits.maxFiles)

package/src/analysis/dead-check.js

@@ -201,7 +201,8 @@ export function computeDead(graph, sources, { entrySet = new Set() } = {}) {
   const deadSymbols = [];
   for (const n of symById.values()) {
     if (isReferenced(n)) continue;
-    const test = isTestFile(n.source_file);
+    // test_surface: extractor-proven test-only symbols (Rust #[cfg(test)]) live in production paths.
+    const test = isTestFile(n.source_file) || n.test_surface === true;
     deadSymbols.push({ id: n.id, file: n.source_file, label: n.label, test, reason: "no inbound edge and name unreferenced outside its file" });
   }
   const deadSet = new Set(deadSymbols.map((s) => s.id));
@@ -209,12 +210,13 @@ export function computeDead(graph, sources, { entrySet = new Set() } = {}) {
   // A production declaration consumed only by tests is live to the raw graph but dead to production.
   // Keep it in a separate review class: this is useful evidence, never an automatic delete verdict.
   const testOnlySymbols = [];
+  const consumerFileOf = (id) => nodesById.get(id)?.source_file || (id.includes("#") ? id.split("#")[0] : id);
+  const isTestConsumer = (id) => nodesById.get(id)?.test_surface === true || isTestFile(consumerFileOf(id));
   for (const n of symById.values()) {
-    if (deadSet.has(n.id) || isTestFile(n.source_file) || isDecorated(n)) continue;
+    if (deadSet.has(n.id) || isTestFile(n.source_file) || n.test_surface === true || isDecorated(n)) continue;
     const sourcesForSymbol = inboundSources.get(String(n.id)) || [];
-    const sourceFiles = sourcesForSymbol.map((id) => nodesById.get(id)?.source_file || (id.includes("#") ? id.split("#")[0] : id));
-    const hasTestInbound = sourceFiles.some((file) => isTestFile(file));
-    const hasProductionInbound = sourceFiles.some((file) => file && !isTestFile(file));
+    const hasTestInbound = sourcesForSymbol.some(isTestConsumer);
+    const hasProductionInbound = sourcesForSymbol.some((id) => consumerFileOf(id) && !isTestConsumer(id));
     const name = bareName(n.label);
     const occurrenceSet = occurrenceFiles.get(name) || new Set();
     const externalOccurrences = [...occurrenceSet].filter((file) => file !== n.source_file);
@@ -232,7 +234,7 @@ export function computeDead(graph, sources, { entrySet = new Set() } = {}) {
       label: n.label,
       test: false,
       reason: "referenced only from test/e2e code; no production consumer was found",
-      testConsumerFiles: [...new Set(sourceFiles.filter((file) => file && isTestFile(file)))].sort(),
+      testConsumerFiles: [...new Set(sourcesForSymbol.filter(isTestConsumer).map(consumerFileOf).filter(Boolean))].sort(),
       evidence: hasTestInbound ? "graph" : hasExactTestInbound ? "exact-semantic" : "lexical",
       publicApi: n.exported === true || ["public", "protected"].includes(String(n.visibility || "").toLowerCase()),
     });

package/src/analysis/dead-code-review.js

@@ -51,6 +51,9 @@ export function computeDeadCodeReview(graph, sources, options = {}) {
   const suppressed = { tests: 0, classified: 0, confidence: 0, path: 0, kind: 0 };
   const candidates = [];
   for (const candidate of raw) {
+    // Node-level test surfaces (Rust #[cfg(test)] symbols in production files) follow the same
+    // include_tests policy as path-classified test files.
+    if (!includeTests && nodesById.get(String(candidate.id))?.test_surface === true) { suppressed.tests += 1; continue; }
     const info = classify(candidate.file, sources.get(candidate.file));
     const allowed = pathAllowed(info, { includeTests, includeClassified });
     if (!allowed.ok) { suppressed[allowed.bucket] += 1; continue; }

package/src/analysis/dep-rules.js

@@ -7,4 +7,4 @@ export {
     globToRe,
     representativeCycle,
 } from './structure/dependency-graph.js'
-export {computeStructureFindings} from './structure/findings.js'
+export {computeStructureFindings, isRustModuleTreeComponent} from './structure/findings.js'

package/src/analysis/git-history/collector.js

@@ -1,4 +1,3 @@
-import {spawn} from 'node:child_process'
 import {isWeavatrixIgnored} from '../../path-ignore.js'
 import {
     boundedHistoryInteger,
@@ -10,48 +9,7 @@ import {
 const HEADER_SEPARATOR = '\x1e'
 const FIELD_SEPARATOR = '\x1f'
 
-export function boundedGitCommand(command, args, options) {
-    return new Promise((resolve, reject) => {
-        const child = spawn(command, args, {
-            cwd: options.cwd, env: options.env, shell: false, windowsHide: true,
-            stdio: ['ignore', 'pipe', 'pipe'],
-        })
-        const stdout = [], stderr = []
-        let stdoutBytes = 0, stderrBytes = 0, truncated = false, timedOut = false, settled = false
-        const finish = (callback) => {
-            if (settled) return
-            settled = true
-            clearTimeout(timer)
-            callback()
-        }
-        const stop = () => { try { child.kill('SIGKILL') } catch { /* process already exited */ } }
-        const timer = setTimeout(() => { timedOut = true; stop() }, options.timeoutMs)
-        child.stdout?.on('data', (chunk) => {
-            if (truncated) return
-            const remaining = options.maxOutputBytes - stdoutBytes
-            if (remaining <= 0) { truncated = true; stop(); return }
-            const kept = chunk.length <= remaining ? chunk : chunk.subarray(0, remaining)
-            stdout.push(kept); stdoutBytes += kept.length
-            if (kept.length !== chunk.length) { truncated = true; stop() }
-        })
-        child.stderr?.on('data', (chunk) => {
-            const remaining = 64 * 1024 - stderrBytes
-            if (remaining <= 0) return
-            const kept = chunk.length <= remaining ? chunk : chunk.subarray(0, remaining)
-            stderr.push(kept); stderrBytes += kept.length
-        })
-        child.on('error', (error) => finish(() => reject(error)))
-        child.on('close', (exitCode) => finish(() => {
-            if (timedOut) return reject(new Error('git history collection timed out'))
-            resolve({
-                stdout: Buffer.concat(stdout),
-                stderr: Buffer.concat(stderr).toString('utf8'),
-                exitCode: Number(exitCode ?? 1),
-                truncated,
-            })
-        }))
-    })
-}
+// git subprocess execution (sync runGit + streaming boundedGitCommand) lives in ../../git-exec.js.
 
 const statNumber = (value) => value === '-'
     ? {value: 0, binary: true}

package/src/analysis/git-history.js

@@ -3,7 +3,8 @@ import {childProcessEnv} from '../child-env.js'
 import {createRepoBoundary} from '../repo-path.js'
 import {loadWeavatrixIgnore} from '../path-ignore.js'
 import {buildGitHistoryAnalytics} from './git-history/analytics.js'
-import {boundedGitCommand, parseGitNumstatLog} from './git-history/collector.js'
+import {parseGitNumstatLog} from './git-history/collector.js'
+import {boundedGitCommand} from '../git-exec.js'
 import {
     boundedHistoryInteger,
     GIT_FORMAT,

package/src/analysis/git-ref-graph.js

@@ -4,18 +4,15 @@
 import {mkdtempSync, mkdirSync, rmSync} from 'node:fs'
 import {join} from 'node:path'
 import {tmpdir} from 'node:os'
-import {spawnSync} from 'node:child_process'
-import {childProcessEnv} from '../child-env.js'
+import {runGit} from '../git-exec.js'
+import {runCommandSync} from '../process.js'
 import {buildInternalGraph} from '../graph/internal-builder.js'
 import {filterGraphForMode} from '../graph/graph-filter.js'
 
 const SAFE_REF = /^(?!-)[A-Za-z0-9][A-Za-z0-9._\/@{}+~^-]{0,199}$/
 
 function git(repoRoot, args, timeout = 15_000) {
-    return spawnSync('git', ['-C', repoRoot, ...args], {
-        encoding: 'utf8', timeout, env: childProcessEnv(), windowsHide: true,
-        maxBuffer: 4 * 1024 * 1024,
-    })
+    return runGit(repoRoot, args, {timeout, maxBuffer: 4 * 1024 * 1024})
 }
 
 export function resolveGitCommit(repoRoot, requestedRef) {
@@ -41,10 +38,7 @@ export async function withGitRefCheckout(repoRoot, requestedRef, operation) {
         if (archived.status !== 0) {
             return {ok: false, error: `git archive failed for ${resolved.ref}: ${String(archived.stderr || '').trim() || 'unknown error'}`}
         }
-        const extracted = spawnSync('tar', ['-xf', archive, '-C', checkout], {
-            encoding: 'utf8', timeout: 60_000, env: childProcessEnv(), windowsHide: true,
-            maxBuffer: 4 * 1024 * 1024,
-        })
+        const extracted = runCommandSync('tar', ['-xf', archive, '-C', checkout], {timeout: 60_000, maxBuffer: 4 * 1024 * 1024})
         if (extracted.status !== 0) {
             return {ok: false, error: `temporary Git archive extraction failed: ${String(extracted.stderr || '').trim() || 'tar unavailable'}`}
         }

package/src/analysis/hot-path-review.js

@@ -128,7 +128,7 @@ export function computeHotPathReview(graph, options = {}) {
         if (!complexity || !file || !String(node?.id || '').includes('#')) continue
         if (scope.path && file !== scope.path && !file.startsWith(`${scope.path}/`)) { excluded.outOfScope++; continue }
         const classification = classifier.explain(file)
-        const test = hasPathClass(classification, 'test', 'e2e')
+        const test = node?.test_surface === true || hasPathClass(classification, 'test', 'e2e')
         const classified = classification.excluded || hasPathClass(classification, ...NON_PRODUCT)
         if (test && options.includeTests !== true) { excluded.tests++; continue }
         if (classified && options.includeClassified !== true) { excluded.classified++; continue }

package/src/analysis/http-contracts/analysis.js

@@ -85,12 +85,18 @@ export function analyzeHttpContracts(input = {}) {
   for (const backend of backends) {
     for (const endpoint of backend.endpoints) {
       const callsites = [];
+      let endpointMismatches = 0;
+      const methodMismatchSites = [];
       for (const client of clients) {
         for (const call of client.calls) {
           if (call.path && !methodMatches(endpoint.method, call.method)) {
             const expected = pathSegments(normalizeHttpContractPath(endpoint.path));
             const actual = pathSegments(call.path);
-            if (routeShapeMatches(expected, actual) || suffixShapeMatch(expected, actual)) methodMismatches++;
+            if (routeShapeMatches(expected, actual) || suffixShapeMatch(expected, actual)) {
+              methodMismatches++;
+              endpointMismatches++;
+              if (methodMismatchSites.length < 3) methodMismatchSites.push({ clientRepo: client.id, file: call.file, line: call.line, method: call.method });
+            }
             continue;
           }
           const match = matchHttpContract(endpoint, call);
@@ -124,6 +130,8 @@ export function analyzeHttpContracts(input = {}) {
         file: normalizeContractFile(endpoint.file) || null,
         line: Number(endpoint.line) || null,
         callsites,
+        methodMismatches: endpointMismatches,
+        methodMismatchSites,
         liveness: externalUseLiveness(callsites, handlerEvidence),
         affected: affectedForEndpoint(callsites, clients, limits),
       });

package/src/analysis/internal-audit/repo-files.js

@@ -1,7 +1,6 @@
 import {readFileSync, readdirSync} from 'node:fs'
 import {join} from 'node:path'
-import {spawnSync} from 'node:child_process'
-import {childProcessEnv} from '../../child-env.js'
+import {runGit} from '../../git-exec.js'
 import {filterWeavatrixIgnored} from '../../path-ignore.js'
 
 export const readText = (path) => {
@@ -29,10 +28,7 @@ const SKIP_DIRS = new Set([
 
 export function listRepoFiles(repoRoot) {
     try {
-        const result = spawnSync('git', ['-C', repoRoot, 'ls-files', '--cached', '--others', '--exclude-standard', '-z'], {
-            encoding: 'utf8', windowsHide: true, timeout: 15_000, maxBuffer: 32 * 1024 * 1024,
-            env: childProcessEnv(),
-        })
+        const result = runGit(repoRoot, ['ls-files', '--cached', '--others', '--exclude-standard', '-z'], {timeout: 15_000, maxBuffer: 32 * 1024 * 1024})
         if (result.status === 0) {
             const files = String(result.stdout || '').split('\0').filter(Boolean).map((file) => file.replace(/\\/g, '/'))
             return filterWeavatrixIgnored(repoRoot, files)

package/src/analysis/structure/findings.js

@@ -63,6 +63,23 @@ function compileTimeFinding(adjacency, component, runtimeComponents, edges) {
     })
 }
 
+// Idiomatic Rust module trees close compile-time SCCs by construction: the parent (mod.rs, lib.rs,
+// main.rs, or a 2018-edition foo.rs) declares `mod child` while children reach back via super::/crate::.
+// Suppress only when every member is a .rs file sitting under one anchor's directory; genuine
+// cross-directory .rs cycles keep their findings.
+export function isRustModuleTreeComponent(component) {
+    if (!component.every((file) => String(file).endsWith('.rs'))) return false
+    return component.some((anchor) => {
+        const path = String(anchor)
+        const slash = path.lastIndexOf('/')
+        const base = path.slice(slash + 1)
+        const anchorDir = ['mod.rs', 'lib.rs', 'main.rs'].includes(base)
+            ? (slash >= 0 ? path.slice(0, slash) : '')
+            : path.slice(0, -'.rs'.length)
+        return anchorDir !== '' && component.every((member) => member === anchor || String(member).startsWith(`${anchorDir}/`))
+    })
+}
+
 export function computeStructureFindings(graph, {rules = {}, entrySet = new Set(), externalImportFiles = new Set()} = {}) {
     const imports = buildFileImportGraph(graph)
     const findings = []
@@ -73,7 +90,9 @@ export function computeStructureFindings(graph, {rules = {}, entrySet = new Set(
 
     const runtimeKeys = new Set(runtimeComponents.map((component) => [...component].sort().join('\0')))
     const allComponents = findSccs(imports.allAdj).sort((a, b) => b.length - a.length)
-    const compileTimeCouplings = allComponents.filter((component) => !runtimeKeys.has([...component].sort().join('\0')))
+    const compileTimeCandidates = allComponents.filter((component) => !runtimeKeys.has([...component].sort().join('\0')))
+    const rustModuleTreeComponents = compileTimeCandidates.filter((component) => isRustModuleTreeComponent(component))
+    const compileTimeCouplings = compileTimeCandidates.filter((component) => !isRustModuleTreeComponent(component))
     for (const component of compileTimeCouplings.slice(0, MAX_CYCLE_FINDINGS)) {
         findings.push(compileTimeFinding(imports.allAdj, component, runtimeComponents, {
             runtime: imports.edges,
@@ -135,6 +154,7 @@ export function computeStructureFindings(graph, {rules = {}, entrySet = new Set(
             largestTypeCoupling: compileTimeCouplings[0]?.length || 0,
             compileTimeCouplings: compileTimeCouplings.length,
             largestCompileTimeCoupling: compileTimeCouplings[0]?.length || 0,
+            rustModuleTreeCouplings: rustModuleTreeComponents.length,
             orphans: findings.filter((finding) => finding.rule === 'orphan-file').length,
             boundaryViolations: violations.length,
         },

package/src/analysis/task-retrieval.js

@@ -64,6 +64,8 @@ export function retrieveTaskContext(g, {
   const pathAllowed = (node) => {
     const file = fileOf(node)
     if (!file || changedFiles.has(file) || includeClassified === true) return true
+    // Node-level test surfaces (Rust #[cfg(test)]) follow the same production-first policy.
+    if (node?.test_surface === true && !requestedClasses.has('test')) return false
     if (!classificationCache.has(file)) classificationCache.set(file, classifier.explain(file, {content: ''}))
     const info = classificationCache.get(file)
     const classes = PATH_CLASS_NAMES.filter((name) => hasPathClass(info, name))

package/src/build-graph.js

@@ -31,6 +31,14 @@ export function defaultPrecisionMode(env = process.env) {
   return env?.WEAVATRIX_PRECISION === "off" ? "off" : "lsp";
 }
 
+// One-line human status for a precision summary (precisionSummary shape) — the graph_stats
+// composition, shared by the build log and open_repo so every surface reads the same fraction.
+export function precisionStatusLine(precision) {
+  const p = precision || {};
+  if (p.state === "NONE") return `Semantic precision: NONE — ${p.reason || "no eligible JavaScript/TypeScript semantic targets"}`;
+  return `Semantic precision: ${p.state || "UNAVAILABLE"} — ${p.queried || 0}/${p.candidates || 0} bounded target(s) queried${p.truncated ? " (truncated)" : ""}; ${p.verifiedEdges || 0} EXACT_LSP edge(s)${p.state !== "COMPLETE" && p.reason ? `; ${p.reason}` : ""}`;
+}
+
 function buildGraphInWorker(payload) {
   return new Promise((resolve, reject) => {
     let worker;
@@ -193,7 +201,7 @@ export async function buildGraphForRepo(repoPath, {
       hotspots: built.hotspots,
       refresh: built.refresh,
       precision: built.precision,
-      log: `built-in builder: ${built.nodes} nodes, ${built.links} static links (${built.refresh?.kind || "full"}: ${built.refresh?.reason || "build"}); semantic precision ${built.precision?.state || "UNAVAILABLE"}, ${built.precision?.verifiedEdges || 0} EXACT_LSP edge(s)`
+      log: `built-in builder: ${built.nodes} nodes, ${built.links} static links (${built.refresh?.kind || "full"}: ${built.refresh?.reason || "build"}); ${precisionStatusLine(built.precision)}`
     };
   } catch (error) {
     return { ok: false, error: `graph build failed: ${error.message}`, builder: "internal" };

package/src/git-exec.js

@@ -0,0 +1,67 @@
+// The single chokepoint for launching git. Every git subprocess in the package routes through here,
+// so there is one place that applies `-C <repoRoot>` containment, a credential-stripped child
+// environment, a hidden Windows window and a bounded timeout — instead of a dozen near-identical
+// spawnSync wrappers scattered across the analysis, graph and MCP layers.
+import { spawnSync, spawn } from "node:child_process";
+import { childProcessEnv } from "./child-env.js";
+
+// Synchronous git. Returns the raw spawnSync result ({status, stdout, stderr, error}) so callers keep
+// their existing status/stdout/error handling. encoding defaults to "utf8" (pass "buffer" for binary
+// stdout); maxBuffer and stdio are forwarded only when provided so Node's defaults are preserved.
+export function runGit(repoRoot, args, options = {}) {
+  const spawnOptions = {
+    encoding: options.encoding ?? "utf8",
+    windowsHide: true,
+    timeout: options.timeout ?? 8000,
+    env: childProcessEnv(options.env),
+  };
+  if (options.maxBuffer != null) spawnOptions.maxBuffer = options.maxBuffer;
+  if (options.stdio != null) spawnOptions.stdio = options.stdio;
+  return spawnSync("git", ["-C", repoRoot, ...args], spawnOptions);
+}
+
+// Streaming git for the history collector: byte-bounded stdout, a hard timeout that SIGKILLs the child,
+// and a resolved {stdout, stderr, exitCode, truncated}. The caller owns cwd/env so reads can be scoped
+// to a specific worktree.
+export function boundedGitCommand(command, args, options) {
+  return new Promise((resolve, reject) => {
+    const child = spawn(command, args, {
+      cwd: options.cwd, env: options.env, shell: false, windowsHide: true,
+      stdio: ["ignore", "pipe", "pipe"],
+    });
+    const stdout = [], stderr = [];
+    let stdoutBytes = 0, stderrBytes = 0, truncated = false, timedOut = false, settled = false;
+    const finish = (callback) => {
+      if (settled) return;
+      settled = true;
+      clearTimeout(timer);
+      callback();
+    };
+    const stop = () => { try { child.kill("SIGKILL"); } catch { /* process already exited */ } };
+    const timer = setTimeout(() => { timedOut = true; stop(); }, options.timeoutMs);
+    child.stdout?.on("data", (chunk) => {
+      if (truncated) return;
+      const remaining = options.maxOutputBytes - stdoutBytes;
+      if (remaining <= 0) { truncated = true; stop(); return; }
+      const kept = chunk.length <= remaining ? chunk : chunk.subarray(0, remaining);
+      stdout.push(kept); stdoutBytes += kept.length;
+      if (kept.length !== chunk.length) { truncated = true; stop(); }
+    });
+    child.stderr?.on("data", (chunk) => {
+      const remaining = 64 * 1024 - stderrBytes;
+      if (remaining <= 0) return;
+      const kept = chunk.length <= remaining ? chunk : chunk.subarray(0, remaining);
+      stderr.push(kept); stderrBytes += kept.length;
+    });
+    child.on("error", (error) => finish(() => reject(error)));
+    child.on("close", (exitCode) => finish(() => {
+      if (timedOut) return reject(new Error("git history collection timed out"));
+      resolve({
+        stdout: Buffer.concat(stdout),
+        stderr: Buffer.concat(stderr).toString("utf8"),
+        exitCode: Number(exitCode ?? 1),
+        truncated,
+      });
+    }));
+  });
+}