etymd 0.11.0 → 0.13.0

package/CHANGELOG.md

@@ -1,5 +1,49 @@
 # etymd
 
+## 0.13.0
+
+### Minor Changes
+
+- 9b45d54: screen: the machine-path check now also skips on upstream-owned files
+
+  The upstream-ownership exemption (0.12.0) skipped vocabulary-class patterns on files
+  byte-identical to public upstream, but the absolute-home-path check still fired on
+  them. Forks of container-based projects carry upstream docs full of generic container
+  paths (the `/home/<user>/...` shape), so a sync kept blocking on paths that are
+  already public in the upstream repo and never name the local machine.
+
+  An upstream-owned file is byte-identical to a public upstream blob, so any path it
+  names is already public and cannot leak this machine. The machine-path check is now
+  skipped for upstream-owned files, exactly like vocabulary-class patterns. Secret-class
+  patterns stay absolute on every file, and the machine-path check still fires on all
+  non-upstream files (the common case) unchanged.
+
+## 0.12.0
+
+### Minor Changes
+
+- 060c50b: screen: forks can exempt upstream-owned files from vocabulary-class patterns
+
+  A fork that syncs from upstream stages hundreds of upstream files; a pattern that
+  matches ordinary wording (a "vocabulary" pattern, e.g. words a leaky _description_
+  would use) then fires on upstream's own already-public docs and blocks the sync
+  commit. This adds:
+
+  - **Pattern classes.** A `# class: vocabulary` directive in the pattern file marks
+    the patterns beneath it as vocabulary; everything else stays `secret`. Files with
+    no directive are unchanged — every pattern absolute.
+  - **Upstream ownership by blob identity.** With `--upstream <remote>` or
+    `git config etymd.upstream <remote>`, a staged/tree file whose content matches any
+    blob at the tips of `refs/remotes/<remote>/*` is upstream-owned, and its
+    vocabulary-class patterns are skipped. Secret-class patterns and the machine-path
+    check stay absolute everywhere. Matching is by blob sha across all upstream refs,
+    so a file copied verbatim to a different path (an adapter lifted from another
+    upstream branch) is still recognised, and a file the fork _modifies_ is screened
+    in full.
+  - **Fails closed and loud.** If the upstream ref cannot be read (a rebuild dropped
+    the remote), nothing is exempted and a one-line notice says so — the config signal
+    persists so the fork still knows it is a fork.
+
 ## 0.11.0
 
 ### Minor Changes

package/dist/{approve-OUJ4UPGK.js → approve-YUT43YLC.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import { section, theme, print, renderBaselineDrift, glyph } from './chunk-5BVKFJWM.js';
-import { scanProject } from './chunk-5GQ52DK6.js';
-import { VERSION } from './chunk-J7XVOTZL.js';
+import { scanProject } from './chunk-YXOAPMQH.js';
+import { VERSION } from './chunk-D3R74TJ2.js';
 import { readBaseline, summarizeBaselineDrift, isDriftEmpty, writeBaseline, deriveProfile } from './chunk-JHZ2BN4U.js';
 import { PACK_VERSION } from './chunk-LSZGCKIQ.js';
 import './chunk-4VPBP6K6.js';

package/dist/{audit-A75FVXHD.js → audit-YRT4SWSQ.js}

@@ -1,12 +1,12 @@
 #!/usr/bin/env node
-export { run } from './chunk-2KVQBL4E.js';
-import './chunk-JVLWM27I.js';
+export { run } from './chunk-DSAQ5S5D.js';
+import './chunk-PXRLOEN5.js';
 import './chunk-F75Q43BC.js';
 import './chunk-3E2IPCRY.js';
 import './chunk-DWL2IKZH.js';
 import './chunk-5BVKFJWM.js';
-import './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
+import './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
 import './chunk-JHZ2BN4U.js';
 import './chunk-LSZGCKIQ.js';
 import './chunk-4VPBP6K6.js';

package/dist/{brief-MYUESUMM.js → brief-Z5S6OY2M.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import { print, theme } from './chunk-5BVKFJWM.js';
-import { scanProject } from './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
+import { scanProject } from './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
 import { ETYMD_DIR } from './chunk-JHZ2BN4U.js';
 import './chunk-LSZGCKIQ.js';
 import './chunk-4VPBP6K6.js';

package/dist/{chunk-J7XVOTZL.js → chunk-D3R74TJ2.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 // package.json
 var package_default = {
-  version: "0.11.0"};
+  version: "0.13.0"};
 
 // src/version.ts
 var VERSION = package_default.version;

package/dist/{chunk-2KVQBL4E.js → chunk-DSAQ5S5D.js}

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { parseFailOnTier, runAudit, meetsFailOn } from './chunk-JVLWM27I.js';
+import { parseFailOnTier, runAudit, meetsFailOn } from './chunk-PXRLOEN5.js';
 import { print, section, theme, renderLensCoverage, renderFindings, renderLedgerDiff } from './chunk-5BVKFJWM.js';
 
 // src/commands/audit.ts

package/dist/{chunk-JVLWM27I.js → chunk-PXRLOEN5.js}

@@ -2,7 +2,7 @@
 import { measureContext } from './chunk-F75Q43BC.js';
 import { readLedger, reconcileLedger, writeLedger, visibleFindings } from './chunk-3E2IPCRY.js';
 import { DEFAULT_CONFIG, readConfig, CONFIG_FILE } from './chunk-DWL2IKZH.js';
-import { scanProject, expandFileGlobs } from './chunk-5GQ52DK6.js';
+import { scanProject, expandFileGlobs } from './chunk-YXOAPMQH.js';
 import { ETYMD_DIR, writeCachedFacts, readBaseline, deriveProfile, baselineCarriesMachinePath, BASELINE_FILE } from './chunk-JHZ2BN4U.js';
 import { PACK_VERSION } from './chunk-LSZGCKIQ.js';
 import { pathExists, readJson, readText, isDirectory, matchesAnyGlob, git, normalizeRelPath, isCiEnvironment, isExecutable } from './chunk-4VPBP6K6.js';

package/dist/{chunk-5GQ52DK6.js → chunk-YXOAPMQH.js}

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { VERSION } from './chunk-J7XVOTZL.js';
+import { VERSION } from './chunk-D3R74TJ2.js';
 import { PACK_VERSION } from './chunk-LSZGCKIQ.js';
 import { normalizeRelPath, isDirectory, readJson, git, matchesAnyGlob, pathExists, readText } from './chunk-4VPBP6K6.js';
 import path from 'path';

package/dist/cli.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { VERSION } from './chunk-J7XVOTZL.js';
+import { VERSION } from './chunk-D3R74TJ2.js';
 import path from 'path';
 import { Command } from 'commander';
 import pc from 'picocolors';
@@ -31,7 +31,7 @@ program.command("audit").description("Verify every instruction claim against the
   "exit non-zero when findings at/above this tier exist (risk|gap|polish)"
 ).action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./audit-A75FVXHD.js');
+    const { run } = await import('./audit-YRT4SWSQ.js');
     await run({
       cwd: resolveCwd(cmd),
       json: opts.json,
@@ -44,7 +44,7 @@ program.command("audit").description("Verify every instruction claim against the
 );
 program.command("init").description("Onboard the truth guard: approve the baseline; scaffold AGENTS.md only if missing").option("-y, --yes", "accept defaults without prompting (never overwrites)").action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./init-QLGIYFCX.js');
+    const { run } = await import('./init-XJEYU2HC.js');
     await run({ cwd: resolveCwd(cmd), yes: opts.yes });
   })
 );
@@ -52,19 +52,19 @@ program.command("approve").description(
   "Re-approve the committed baseline after intentional structural changes (non-interactive)"
 ).action(
   (_opts, cmd) => action(async () => {
-    const { run } = await import('./approve-OUJ4UPGK.js');
+    const { run } = await import('./approve-YUT43YLC.js');
     await run({ cwd: resolveCwd(cmd) });
   })
 );
 program.command("scan").description("Deterministically reckon the project into a facts index").option("--json", "print the raw facts as JSON").option("--no-save", "do not write the .etymd cache").action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./scan-VWEXWM6G.js');
+    const { run } = await import('./scan-B24RZAV2.js');
     await run({ cwd: resolveCwd(cmd), json: opts.json, save: opts.save });
   })
 );
 program.command("doctor").description('Alias for `audit --truth` \u2014 "are the recorded instructions still true?"').option("--json", "print findings as JSON").action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./doctor-OZQTOUFR.js');
+    const { run } = await import('./doctor-7YKDXSKM.js');
     await run({ cwd: resolveCwd(cmd), json: opts.json });
   })
 );
@@ -76,7 +76,7 @@ program.command("context").description("Measure the always-loaded context footpr
 );
 program.command("brief").description("Emit a grounded briefing for the in-repo agent to complete the semantic layer").option("--human", "write a human onboarding brief instead of the agent briefing").action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./brief-MYUESUMM.js');
+    const { run } = await import('./brief-Z5S6OY2M.js');
     await run({ cwd: resolveCwd(cmd), human: opts.human });
   })
 );
@@ -111,7 +111,7 @@ var fleet = program.command("fleet").description(
   "exit non-zero when findings at/above this tier exist (risk|gap|polish)"
 ).action(
   (opts, cmd) => action(async () => {
-    const { sweep } = await import('./fleet-Z7W3QBEV.js');
+    const { sweep } = await import('./fleet-YQ35KGEP.js');
     await sweep({
       cwd: resolveCwd(cmd),
       manifest: opts.manifest,
@@ -126,7 +126,7 @@ var fleet = program.command("fleet").description(
 );
 fleet.command("check").description("Validate the manifest pair only \u2014 no lenses: dangling mappings, duplicates, leaks").option("--manifest <file>", "the fleet manifest \u2014 required unless the cwd holds registry.json").option("--json", "print the findings as JSON (EXPERIMENTAL through 0.2.x)").action(
   (_opts, cmd) => action(async () => {
-    const { check } = await import('./fleet-Z7W3QBEV.js');
+    const { check } = await import('./fleet-YQ35KGEP.js');
     const opts = cmd.optsWithGlobals();
     await check({ cwd: resolveCwd(cmd), manifest: opts.manifest, json: opts.json });
   })
@@ -136,7 +136,7 @@ fleet.command("add").argument("<dir>", "directory of the project to register").d
   "public-repo | public-bound | private \u2014 mandatory for personal entries"
 ).option("-y, --yes", "skip prompts; every mandatory value must be passed as a flag").option("--manifest <file>", "the fleet manifest \u2014 required unless the cwd holds registry.json").action(
   (dir, opts, cmd) => action(async () => {
-    const { add } = await import('./fleet-Z7W3QBEV.js');
+    const { add } = await import('./fleet-YQ35KGEP.js');
     const shared = cmd.optsWithGlobals();
     await add({
       cwd: resolveCwd(cmd),
@@ -159,7 +159,7 @@ fleet.command("dismiss").argument("<name>", "the registered project name").argum
   "Dismiss a project's finding from any cwd \u2014 corp ledgers persist beside the manifest"
 ).requiredOption("--reason <text>", "why it is dismissed \u2014 recorded so the decision survives").option("--manifest <file>", "the fleet manifest \u2014 required unless the cwd holds registry.json").action(
   (name, id, _opts, cmd) => action(async () => {
-    const { dismiss } = await import('./fleet-Z7W3QBEV.js');
+    const { dismiss } = await import('./fleet-YQ35KGEP.js');
     const opts = cmd.optsWithGlobals();
     await dismiss({
       cwd: resolveCwd(cmd),
@@ -174,29 +174,33 @@ fleet.command("accept").argument("<name>", "the registered project name").argume
   "Accept a project's finding as a known trade-off \u2014 corp ledgers persist beside the manifest"
 ).option("--reason <text>", "optional note on why the trade-off is accepted").option("--manifest <file>", "the fleet manifest \u2014 required unless the cwd holds registry.json").action(
   (name, id, _opts, cmd) => action(async () => {
-    const { accept } = await import('./fleet-Z7W3QBEV.js');
+    const { accept } = await import('./fleet-YQ35KGEP.js');
     const opts = cmd.optsWithGlobals();
     await accept({ cwd: resolveCwd(cmd), manifest: opts.manifest, name, id, reason: opts.reason });
   })
 );
 program.command("screen").description(
   "Content screen: check for text that must never be published. Bring your own patterns \u2014 etymd ships none."
-).option("--staged", "screen what a commit is about to add (default)").option("--tree", "screen every tracked file \u2014 what is about to leave the machine").option("--message <file>", "screen a commit message (the commit-msg hook's $1)").option("--dir <dir>", "screen an unpacked build artifact \u2014 what actually ships").option("--patterns <file>", "pattern file (default: ~/.config/etymd/screen-patterns)").option("--advisory", "report without failing").action(
+).option("--staged", "screen what a commit is about to add (default)").option("--tree", "screen every tracked file \u2014 what is about to leave the machine").option("--message <file>", "screen a commit message (the commit-msg hook's $1)").option("--dir <dir>", "screen an unpacked build artifact \u2014 what actually ships").option("--patterns <file>", "pattern file (default: ~/.config/etymd/screen-patterns)").option("--advisory", "report without failing").option(
+  "--upstream <remote>",
+  "fork's upstream remote: files byte-identical to its refs skip vocabulary-class patterns (secret patterns stay absolute)"
+).action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./screen-NSWNC7VQ.js');
+    const { run } = await import('./screen-CS6PSG7U.js');
     const scope = opts.message ? "message" : opts.dir ? "dir" : opts.tree ? "tree" : "staged";
     await run({
       cwd: resolveCwd(cmd),
       scope,
       target: opts.message ?? opts.dir,
       patterns: opts.patterns,
-      advisory: opts.advisory
+      advisory: opts.advisory,
+      upstream: opts.upstream
     });
   })
 );
 program.command("gates").description("Install the local git-hook gates (process \u2192 pre-commit, correctness \u2192 pre-push)").option("--ci", "note about the CI review gate (ships later; local gates install now)").option("-y, --yes", "skip prompts; never overwrites a hand-edited hook").action(
   (opts, cmd) => action(async () => {
-    const { run } = await import('./gates-47YTMMLZ.js');
+    const { run } = await import('./gates-DU6SDH2M.js');
     await run({ cwd: resolveCwd(cmd), ci: opts.ci, yes: opts.yes });
   })
 );

package/dist/{doctor-OZQTOUFR.js → doctor-7YKDXSKM.js}

@@ -1,12 +1,12 @@
 #!/usr/bin/env node
-import { run } from './chunk-2KVQBL4E.js';
-import './chunk-JVLWM27I.js';
+import { run } from './chunk-DSAQ5S5D.js';
+import './chunk-PXRLOEN5.js';
 import './chunk-F75Q43BC.js';
 import './chunk-3E2IPCRY.js';
 import './chunk-DWL2IKZH.js';
 import './chunk-5BVKFJWM.js';
-import './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
+import './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
 import './chunk-JHZ2BN4U.js';
 import './chunk-LSZGCKIQ.js';
 import './chunk-4VPBP6K6.js';

package/dist/{fleet-Z7W3QBEV.js → fleet-YQ35KGEP.js}

@@ -1,12 +1,12 @@
 #!/usr/bin/env node
 import { dismiss, accept } from './chunk-2VLNI3L2.js';
-import { parseFailOnTier, meetsFailOn, runAudit } from './chunk-JVLWM27I.js';
+import { parseFailOnTier, meetsFailOn, runAudit } from './chunk-PXRLOEN5.js';
 import './chunk-F75Q43BC.js';
 import { readLedger } from './chunk-3E2IPCRY.js';
 import { DEFAULT_CONFIG } from './chunk-DWL2IKZH.js';
 import { print, theme, section, renderFleetRows, renderFleetNotes, renderFindings, TIER_BADGE, glyph } from './chunk-5BVKFJWM.js';
-import { scanProject } from './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
+import { scanProject } from './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
 import { ETYMD_DIR } from './chunk-JHZ2BN4U.js';
 import './chunk-LSZGCKIQ.js';
 import { readText, pathExists, git, isDirectory } from './chunk-4VPBP6K6.js';
@@ -688,7 +688,7 @@ async function checkCorpEmails(manifest, findings, disclosures) {
 }
 async function checkGateDrift(manifest, findings, disclosures) {
   const { planWorkflow } = await import('./generate-PMCP37DC.js');
-  const { scanProject: scanProject2 } = await import('./scan-W7RHZVSF.js');
+  const { scanProject: scanProject2 } = await import('./scan-QDNN7ER3.js');
   const { readConfig } = await import('./config-XAH6PA5G.js');
   for (const entry of manifest.entries) {
     const root = entry.resolvedRoot;

package/dist/{gates-47YTMMLZ.js → gates-DU6SDH2M.js}

@@ -3,8 +3,8 @@ import { applyFiles } from './chunk-GWGKEPRX.js';
 import { deriveFailOn, derivedCommands, planWorkflow, runPrefix, isSafeGateCommand } from './chunk-HRJJQCMT.js';
 import { readConfig, CONFIG_FILE, configPath } from './chunk-DWL2IKZH.js';
 import { section, print, glyph, theme, renderPlan } from './chunk-5BVKFJWM.js';
-import { scanProject } from './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
+import { scanProject } from './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
 import './chunk-JHZ2BN4U.js';
 import './chunk-LSZGCKIQ.js';
 import { readText, git } from './chunk-4VPBP6K6.js';

package/dist/index.js

@@ -30,7 +30,7 @@ var init_package = __esm({
   "package.json"() {
     package_default = {
       name: "etymd",
-      version: "0.11.0",
+      version: "0.13.0",
       description: "Keep your agent instructions true \u2014 verify AGENTS.md, CLAUDE.md, rules & skills against the actual repo, with drift caught over time and a regression ledger.",
       keywords: [
         "cli",

package/dist/{init-QLGIYFCX.js → init-XJEYU2HC.js}

@@ -3,8 +3,8 @@ import { applyFiles } from './chunk-GWGKEPRX.js';
 import { planWorkflow } from './chunk-HRJJQCMT.js';
 import './chunk-DWL2IKZH.js';
 import { theme, renderFacts, print, renderPlan, section, glyph } from './chunk-5BVKFJWM.js';
-import { scanProject } from './chunk-5GQ52DK6.js';
-import { VERSION } from './chunk-J7XVOTZL.js';
+import { scanProject } from './chunk-YXOAPMQH.js';
+import { VERSION } from './chunk-D3R74TJ2.js';
 import { writeCachedFacts, deriveProfile, writeBaseline, CACHE_DIR } from './chunk-JHZ2BN4U.js';
 import { PACK_VERSION } from './chunk-LSZGCKIQ.js';
 import { git, readText } from './chunk-4VPBP6K6.js';

package/dist/{scan-VWEXWM6G.js → scan-B24RZAV2.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import { print, renderFacts, theme } from './chunk-5BVKFJWM.js';
-import { scanProject } from './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
+import { scanProject } from './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
 import { writeCachedFacts } from './chunk-JHZ2BN4U.js';
 import './chunk-LSZGCKIQ.js';
 import './chunk-4VPBP6K6.js';

package/dist/scan-QDNN7ER3.js

@@ -0,0 +1,5 @@
+#!/usr/bin/env node
+export { scanProject } from './chunk-YXOAPMQH.js';
+import './chunk-D3R74TJ2.js';
+import './chunk-LSZGCKIQ.js';
+import './chunk-4VPBP6K6.js';

package/dist/{screen-NSWNC7VQ.js → screen-CS6PSG7U.js}

@@ -4,6 +4,7 @@ import { git, readText, pathExists } from './chunk-4VPBP6K6.js';
 import { promises } from 'fs';
 import path from 'path';
 
+var CLASS_DIRECTIVE = /^#[ \t]*class:[ \t]*(secret|vocabulary)\b/i;
 var MACHINE_PATH_RE = /\/(?:Users|home)\/[A-Za-z0-9._-]+\//;
 var ALLOW_MARKER = "allow-published-string";
 var ALLOW_FILE = ".etymd-screen-allow";
@@ -48,14 +49,28 @@ function makeAllowEntry(patternStr) {
     };
   }
 }
+function compileOnePattern(l) {
+  try {
+    return new RegExp(l, "i");
+  } catch {
+    return new RegExp(l.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"), "i");
+  }
+}
 function compilePatterns(raw) {
-  return raw.split("\n").map((l) => l.trim()).filter((l) => l && !l.startsWith("#")).map((l) => {
-    try {
-      return new RegExp(l, "i");
-    } catch {
-      return new RegExp(l.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"), "i");
+  const out = [];
+  let cls = "secret";
+  for (const rawLine of raw.split("\n")) {
+    const l = rawLine.trim();
+    if (!l) continue;
+    const dir = CLASS_DIRECTIVE.exec(l);
+    if (dir) {
+      cls = dir[1].toLowerCase();
+      continue;
     }
-  });
+    if (l.startsWith("#")) continue;
+    out.push({ re: compileOnePattern(l), cls });
+  }
+  return out;
 }
 var META = ".[]*^$+?(){}|";
 function patternLiteral(source) {
@@ -101,19 +116,23 @@ async function selfIdentities(cwd) {
   names.delete("");
   return [...names];
 }
-function screenText(text, file, patterns, allow = []) {
+function screenText(text, file, patterns, allow = [], upstreamOwned = false) {
+  const classed = patterns.map(
+    (p) => p instanceof RegExp ? { re: p, cls: "secret" } : p
+  );
+  const active = upstreamOwned ? classed.filter((p) => p.cls === "secret") : classed;
   const hits = [];
   const lines = text.split("\n");
   for (const [i, line] of lines.entries()) {
     if (line.includes(ALLOW_MARKER)) continue;
     if (allow.some((entry) => entry.pattern.test(line))) continue;
-    for (const re of patterns) {
+    for (const { re } of active) {
       if (re.test(line)) {
         hits.push({ file, line: i + 1, text: line.trim().slice(0, 160), reason: String(re) });
         break;
       }
     }
-    if (MACHINE_PATH_RE.test(line)) {
+    if (!upstreamOwned && MACHINE_PATH_RE.test(line)) {
       hits.push({
         file,
         line: i + 1,
@@ -124,6 +143,30 @@ function screenText(text, file, patterns, allow = []) {
   }
   return hits;
 }
+async function collectUpstreamBlobs(cwd, remote) {
+  const refsRaw = await git(cwd, ["for-each-ref", "--format=%(refname)", `refs/remotes/${remote}/`]);
+  const refs = (refsRaw ?? "").split("\n").filter(Boolean);
+  if (!refs.length) return null;
+  const blobs = /* @__PURE__ */ new Set();
+  for (const ref of refs) {
+    const tree = await git(cwd, ["ls-tree", "-r", ref]);
+    if (tree === null) continue;
+    for (const line of tree.split("\n")) {
+      const m = /^\d+ blob ([0-9a-f]+)\t/.exec(line);
+      if (m) blobs.add(m[1]);
+    }
+  }
+  return blobs.size ? blobs : null;
+}
+async function stagedBlobShas(cwd) {
+  const raw = await git(cwd, ["ls-files", "-s"]);
+  const map = /* @__PURE__ */ new Map();
+  for (const line of (raw ?? "").split("\n")) {
+    const m = /^\d+ ([0-9a-f]+) \d+\t(.+)$/.exec(line);
+    if (m) map.set(m[2], m[1]);
+  }
+  return map;
+}
 async function walk(dir, out, depth = 0) {
   if (depth > 12) return;
   let entries;
@@ -189,7 +232,25 @@ async function run(opts) {
     return true;
   });
   const allow = validatedEntries;
-  const active = selves.length ? patterns.filter((re) => !isSelfName(re, selves)) : patterns;
+  const active = selves.length ? patterns.filter((p) => !isSelfName(p.re, selves)) : patterns;
+  const hasVocab = active.some((p) => p.cls === "vocabulary");
+  const upstreamRemote = hasVocab && (opts.scope === "staged" || opts.scope === "tree") ? opts.upstream ?? await git(opts.cwd, ["config", "--get", "etymd.upstream"]) ?? void 0 : void 0;
+  let upstreamBlobs = null;
+  let blobByPath = /* @__PURE__ */ new Map();
+  let upstreamNote = "";
+  if (upstreamRemote) {
+    upstreamBlobs = await collectUpstreamBlobs(opts.cwd, upstreamRemote);
+    if (upstreamBlobs === null) {
+      upstreamNote = `upstream ref '${upstreamRemote}' unavailable \u2014 vocabulary exemption off, screening every pattern`;
+    } else {
+      blobByPath = await stagedBlobShas(opts.cwd);
+    }
+  }
+  const isUpstreamOwned = (rel) => {
+    if (!upstreamBlobs) return false;
+    const sha = blobByPath.get(rel);
+    return !!sha && upstreamBlobs.has(sha);
+  };
   if (usingLegacy) {
     print(
       `  ${glyph.partial} ${theme.warn(`Using legacy ${LEGACY_ALLOW_FILE} file \u2014 rename to ${ALLOW_FILE} and add provenance`)}`
@@ -238,9 +299,12 @@ async function run(opts) {
         continue;
       }
       scanned++;
-      hits.push(...screenText(raw, rel, active, allow));
+      hits.push(...screenText(raw, rel, active, allow, isUpstreamOwned(rel)));
     }
   }
+  if (upstreamNote) {
+    print(`  ${glyph.partial} ${theme.warn(upstreamNote)}`);
+  }
   section(
     `Content screen ${theme.dim(
       `\xB7 ${opts.scope} \xB7 ${scanned} file(s)${skippedBinary ? ` \xB7 ${skippedBinary} binary skipped` : ""}`
@@ -264,4 +328,4 @@ async function run(opts) {
   if (!opts.advisory) process.exitCode = 1;
 }
 
-export { compileAllow, isSelfName, patternLiteral, readScreenable, run, screenText, selfIdentities };
+export { collectUpstreamBlobs, compileAllow, compilePatterns, isSelfName, patternLiteral, readScreenable, run, screenText, selfIdentities, stagedBlobShas };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "etymd",
-  "version": "0.11.0",
+  "version": "0.13.0",
   "description": "Keep your agent instructions true — verify AGENTS.md, CLAUDE.md, rules & skills against the actual repo, with drift caught over time and a regression ledger.",
   "keywords": [
     "cli",

package/dist/scan-W7RHZVSF.js

@@ -1,5 +0,0 @@
-#!/usr/bin/env node
-export { scanProject } from './chunk-5GQ52DK6.js';
-import './chunk-J7XVOTZL.js';
-import './chunk-LSZGCKIQ.js';
-import './chunk-4VPBP6K6.js';