deps: switch Amp package to @ampcode/cli

.github/workflows/regenerate-prompts.yml

@@ -40,13 +40,13 @@ jobs:
         run: |
           git config user.name "github-actions[bot]"
           git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
-          git add -A prompts amp-prompts.md
+          git add -A README.md prompts subagents skills
           if git diff --cached --quiet; then
             echo "No prompt changes."
             exit 0
           fi
-          AMP_VERSION=$(bun -e "console.log(require('./node_modules/@sourcegraph/amp/package.json').version)")
-          git commit -m "Regenerate prompts for @sourcegraph/amp@${AMP_VERSION}"
+          AMP_VERSION=$(bun -e "console.log(require('./node_modules/@ampcode/cli/package.json').version)")
+          git commit -m "Regenerate prompts for @ampcode/cli@${AMP_VERSION}"
           git push
 
       - name: Auto-approve amp bump PR

.github/workflows/update-amp.yml

@@ -23,11 +23,11 @@ jobs:
         with:
           bun-version: latest
 
-      - name: Update @sourcegraph/amp
+      - name: Update @ampcode/cli
         id: amp
         run: |
-          current=$(bun -e "console.log(require('./package.json').dependencies['@sourcegraph/amp'])")
-          latest=$(bun pm view @sourcegraph/amp version)
+          current=$(bun -e "console.log(require('./package.json').dependencies['@ampcode/cli'])")
+          latest=$(bun pm view @ampcode/cli version)
 
           echo "current=${current}" >> "$GITHUB_OUTPUT"
           echo "latest=${latest}" >> "$GITHUB_OUTPUT"
@@ -37,7 +37,7 @@ jobs:
             exit 0
           fi
 
-          bun add "@sourcegraph/amp@${latest}"
+          bun add "@ampcode/cli@${latest}"
           bun run extract
           echo "changed=true" >> "$GITHUB_OUTPUT"
 
@@ -49,10 +49,10 @@ jobs:
           token: ${{ secrets.GITHUB_TOKEN }}
           branch: bot/update-amp
           delete-branch: true
-          title: "deps: update @sourcegraph/amp to ${{ steps.amp.outputs.latest }}"
-          commit-message: "deps: update @sourcegraph/amp to ${{ steps.amp.outputs.latest }}"
+          title: "deps: update @ampcode/cli to ${{ steps.amp.outputs.latest }}"
+          commit-message: "deps: update @ampcode/cli to ${{ steps.amp.outputs.latest }}"
           body: |
-            Updates `@sourcegraph/amp` from `${{ steps.amp.outputs.current }}` to `${{ steps.amp.outputs.latest }}`.
+            Updates `@ampcode/cli` from `${{ steps.amp.outputs.current }}` to `${{ steps.amp.outputs.latest }}`.
 
             This workflow also runs `bun run extract` so the generated prompt files stay in sync with the package bump.
           labels: amp-bump

README.md

@@ -1,19 +1,25 @@
 # amp-prompts
 
 Tooling that extracts the system prompts baked into the
-[`@sourcegraph/amp`](https://www.npmjs.com/package/@sourcegraph/amp) CLI bundle and
-writes human-readable Markdown files under `prompts/` and `skills/`.
-
-The Amp bundle ships as a single minified `dist/main.js`. Each prompt is a
-template literal that often takes feature-flag arguments (e.g. `enableOracle`,
-`enableDiagnostics`) and references other minified identifiers like `R4`, `V6`,
-`AA`. This repo:
-
-1. Pins `@sourcegraph/amp` as a normal dependency so Dependabot can open a PR
+[`@ampcode/cli`](https://www.npmjs.com/package/@ampcode/cli) CLI and
+writes human-readable Markdown files under `prompts/`, `subagents/`, and
+`skills/`.
+
+Older `@sourcegraph/amp` packages shipped a single minified `dist/main.js`.
+Amp now ships through `@ampcode/cli`, whose platform packages contain a native
+executable with the bundled JavaScript embedded inside it. The extractor
+supports both layouts: it reads the legacy `dist/main.js` when present,
+otherwise scans the installed native Amp binary for the embedded bundle. Each
+prompt is a template literal that often takes feature-flag arguments (e.g.
+`enableOracle`, `enableDiagnostics`) and references other minified identifiers
+like `R4`, `V6`, `AA`. This repo:
+
+1. Pins `@ampcode/cli` as a normal dependency so Dependabot can open a PR
    each time a new bundle is published.
 2. Parses the installed bundle with `@babel/parser`, locates every arrow
-   function or template literal whose body looks like a prompt, and
-   evaluates it inside a `node:vm` sandbox.
+   function or template literal whose body looks like a prompt, rejects
+   generated runtime bundles and trivial guard strings, and evaluates each
+   candidate inside a `node:vm` sandbox.
 3. The sandbox's globals are a `Proxy` whose backing map is built from every
    `name = "value"` assignment in the bundle, so a reference to `R4`
    resolves to the string `"finder"`, `V6` to `"Read"`, and so on. Any
@@ -55,8 +61,8 @@ The script writes:
 
 <!-- BEGIN GENERATED CATALOG -->
 
-Source: node_modules/@sourcegraph/amp/dist/main.js
-Package: @sourcegraph/amp@0.0.1781003299-g2fde29
+Source: node_modules/@ampcode/cli-darwin-arm64/amp#embedded-js@62734374
+Package: @ampcode/cli@0.0.1781102632-gaaab69
 
 Notes:
 - Extracted by parsing the bundle with `@babel/parser`, locating prompt-producing arrow functions and template literals, then evaluating each in a `node:vm` sandbox.
@@ -66,23 +72,22 @@ Notes:
 
 ## Prompts
 
-- [prompt](prompts/prompt.md) — line 744
-- [subagent-summary](prompts/subagent-summary.md) — line 1527
-- [ai-assistant](prompts/ai-assistant.md) — line 2483
+- [ai-assistant](prompts/ai-assistant.md) — line 2211
+- [subagent-summary](prompts/subagent-summary.md) — line 4476
 
 ## Subagents
 
-- [review](subagents/review.md) — line 1730
+- [review](subagents/review.md) — line 1933
 
 ## Skills
 
-- [code-review-skill](skills/code-review-skill.md) — line 1040
+- [code-review-skill](skills/code-review-skill.md) — line 1804
 
 <!-- END GENERATED CATALOG -->
 
 ## Bumping Amp
 
-`@sourcegraph/amp` is pinned exactly in `package.json`. The workflow at
+`@ampcode/cli` is pinned exactly in `package.json`. The workflow at
 `.github/workflows/update-amp.yml` runs daily, can also be triggered manually,
 asks the npm registry for the latest published version with `bun pm view`,
 updates the exact pin with `bun add`, runs `bun run extract`, and opens a PR
@@ -99,7 +104,7 @@ also enables auto-merge for `amp-bump` PRs opened by `github-actions[bot]` or
 
 ```
 extract-amp-prompts.mjs       # the extractor
-package.json                  # pins @sourcegraph/amp
+package.json                  # pins @ampcode/cli
 .github/workflows/update-amp.yml
 .github/workflows/regenerate-prompts.yml
 prompts/                      # generated; refreshed by the workflow

extract-amp-prompts.mjs

@@ -5,18 +5,112 @@ import vm from 'node:vm';
 import { parse } from '@babel/parser';
 
 const ROOT = path.dirname(url.fileURLToPath(import.meta.url));
-const AMP_PKG_DIR = path.join(ROOT, 'node_modules', '@sourcegraph', 'amp');
-const BUNDLE = path.join(AMP_PKG_DIR, 'dist', 'main.js');
-const PKG = path.join(AMP_PKG_DIR, 'package.json');
+const AMPCODE_CLI_DIR = path.join(ROOT, 'node_modules', '@ampcode', 'cli');
+const AMPCODE_CLI_PKG = path.join(AMPCODE_CLI_DIR, 'package.json');
+const LEGACY_SOURCEGRAPH_DIR = path.join(ROOT, 'node_modules', '@sourcegraph', 'amp');
+const LEGACY_SOURCEGRAPH_PKG = path.join(LEGACY_SOURCEGRAPH_DIR, 'package.json');
+const LEGACY_BUNDLE = path.join(LEGACY_SOURCEGRAPH_DIR, 'dist', 'main.js');
 const PROMPTS_DIR = path.join(ROOT, 'prompts');
 const SUBAGENTS_DIR = path.join(ROOT, 'subagents');
 const SKILLS_DIR = path.join(ROOT, 'skills');
 const README_FILE = path.join(ROOT, 'README.md');
+const MAX_PROMPT_CHARS = 80_000;
 
-const source = fs.readFileSync(BUNDLE, 'utf8');
-const pkg = JSON.parse(fs.readFileSync(PKG, 'utf8'));
+function relative(p) {
+  return path.relative(ROOT, p);
+}
+
+function resolveNativeAmpBinary() {
+  if (!fs.existsSync(AMPCODE_CLI_PKG)) {
+    throw new Error('Could not find @ampcode/cli. Run bun install first.');
+  }
+  const cliPkg = JSON.parse(fs.readFileSync(path.join(AMPCODE_CLI_DIR, 'package.json'), 'utf8'));
+  const platformPackageByKey = {
+    'darwin-arm64': '@ampcode/cli-darwin-arm64',
+    'darwin-x64': '@ampcode/cli-darwin-x64',
+    'linux-arm64': '@ampcode/cli-linux-arm64',
+    'linux-x64': '@ampcode/cli-linux-x64',
+    'win32-x64': '@ampcode/cli-win32-x64',
+  };
+  const key = `${process.platform}-${process.arch}`;
+  const platformPackage = platformPackageByKey[key];
+  if (platformPackage) {
+    const platformDir = path.join(ROOT, 'node_modules', ...platformPackage.split('/'));
+    const binary = path.join(platformDir, process.platform === 'win32' ? 'amp.exe' : 'amp');
+    if (fs.existsSync(binary)) return binary;
+  }
+
+  const binRel = cliPkg.bin?.amp;
+  if (binRel) {
+    const binary = path.join(AMPCODE_CLI_DIR, binRel);
+    if (fs.existsSync(binary) && fs.statSync(binary).size > 100_000) return binary;
+  }
 
-console.log('Parsing bundle...');
+  throw new Error(`Could not find native Amp binary for ${key}. Install @ampcode/cli first.`);
+}
+
+function sourceLikeByte(byte) {
+  return byte === 0x09 || byte === 0x0a || byte === 0x0d || (byte >= 0x20 && byte <= 0x7e) || byte >= 0x80;
+}
+
+function extractEmbeddedBundle(binaryPath) {
+  const buffer = fs.readFileSync(binaryPath);
+  const bundleMarkers = [
+    'Code Review Skill',
+    'Additional instructions from the user',
+    'Please analyze this codebase',
+    'AGENTS.md guidance',
+    'promptFragments',
+    'systemPrompt',
+  ];
+  const candidates = [];
+  let start = -1;
+  for (let i = 0; i <= buffer.length; i += 1) {
+    const isSourceLike = i < buffer.length && sourceLikeByte(buffer[i]);
+    if (isSourceLike && start === -1) start = i;
+    if ((!isSourceLike || i === buffer.length) && start !== -1) {
+      const length = i - start;
+      if (length >= 500_000) {
+        const text = buffer.subarray(start, i).toString('utf8');
+        const markerCount = bundleMarkers.filter((marker) => text.includes(marker)).length;
+        if (text.startsWith('// @bun') && markerCount > 0) {
+          candidates.push({ text, offset: start, length, markerCount });
+        }
+      }
+      start = -1;
+    }
+  }
+
+  candidates.sort((a, b) => b.markerCount - a.markerCount || b.length - a.length);
+  if (candidates.length === 0) {
+    throw new Error(`Could not find embedded Amp JS bundle in ${binaryPath}`);
+  }
+  return candidates[0];
+}
+
+function loadAmpSource() {
+  if (fs.existsSync(LEGACY_BUNDLE)) {
+    const pkg = JSON.parse(fs.readFileSync(LEGACY_SOURCEGRAPH_PKG, 'utf8'));
+    return {
+      source: fs.readFileSync(LEGACY_BUNDLE, 'utf8'),
+      sourceLabel: relative(LEGACY_BUNDLE),
+      pkg,
+    };
+  }
+
+  const pkg = JSON.parse(fs.readFileSync(AMPCODE_CLI_PKG, 'utf8'));
+  const binary = resolveNativeAmpBinary();
+  const embedded = extractEmbeddedBundle(binary);
+  return {
+    source: embedded.text,
+    sourceLabel: `${relative(binary)}#embedded-js@${embedded.offset}`,
+    pkg,
+  };
+}
+
+const { source, sourceLabel, pkg } = loadAmpSource();
+
+console.log(`Parsing ${sourceLabel}...`);
 const ast = parse(source, {
   sourceType: 'unambiguous',
   errorRecovery: true,
@@ -70,7 +164,21 @@ console.log(`Collected ${identMap.size} identifier candidates.`);
 //   1. Arrow/Function expressions whose body is (or returns) a template literal containing prompt-y text.
 //   2. Top-level template literals containing prompt-y text (already-rendered prompts).
 const PROMPT_RE = /You are|Please analyze|AGENTS\.md guidance|MUST answer|fast, parallel code search agent|Librarian|REPL operator|code review|Additional instructions from the user/i;
-const STARTS_LIKE_PROMPT = /^(#|You are|Please analyze|AGENTS\.md guidance|MUST answer|The following|Run comprehensive|Remember:|Additional instructions from the user)/i;
+const STARTS_LIKE_PROMPT = /^(# Code Review Skill\b|#{1,3}\s+.+\bSkill\b|You are|Please analyze|AGENTS\.md guidance|MUST answer|The following|Run comprehensive|Remember:|Additional instructions from the user)/i;
+
+function isBundleLikeText(text) {
+  const trimmed = text.trimStart();
+  if (trimmed.length > MAX_PROMPT_CHARS) return true;
+  if (trimmed.startsWith('#!/usr/bin/env bun')) return true;
+  if (trimmed.startsWith('// @bun\nvar ')) return true;
+  if (/^(var|function)\s+[A-Za-z_$][\w$]*=.*Object\.create/.test(trimmed.slice(0, 300))) return true;
+  return false;
+}
+
+function looksLikePromptText(text) {
+  const trimmed = text.trim();
+  return !isBundleLikeText(trimmed) && PROMPT_RE.test(trimmed) && STARTS_LIKE_PROMPT.test(trimmed);
+}
 
 function lineOf(node) {
   return node.loc?.start?.line ?? 0;
@@ -104,7 +212,7 @@ function walkWithParent(node, parent) {
     }
     if (tpl) {
       const sample = templateLiteralText(tpl);
-      if (PROMPT_RE.test(sample) && STARTS_LIKE_PROMPT.test(sample.trim())) {
+      if (looksLikePromptText(sample)) {
         candidates.push({
           kind: 'function',
           node,
@@ -121,7 +229,7 @@ function walkWithParent(node, parent) {
   // Case B: a bare TemplateLiteral that looks like a prompt.
   if (node.type === 'TemplateLiteral' && (parent?.type !== 'ReturnStatement' && parent?.type !== 'ArrowFunctionExpression')) {
     const sample = templateLiteralText(node);
-    if (PROMPT_RE.test(sample) && STARTS_LIKE_PROMPT.test(sample.trim())) {
+    if (looksLikePromptText(sample)) {
       candidates.push({
         kind: 'template',
         node,
@@ -154,11 +262,11 @@ const PUBLIC_BASE_PROMPT_NAMES = new Map([
   ['frontier', null],
 ]);
 const basePromptNameBySymbol = new Map();
-for (const match of source.matchAll(/case"([^"]+)":I=([A-Za-z_$][\w$]*)\(/g)) {
+for (const match of source.matchAll(/case"([^"]+)":[A-Za-z_$][\w$]*=([A-Za-z_$][\w$]*)\(/g)) {
   const name = PUBLIC_BASE_PROMPT_NAMES.has(match[1]) ? PUBLIC_BASE_PROMPT_NAMES.get(match[1]) : match[1];
   if (name) basePromptNameBySymbol.set(match[2], name);
 }
-const defaultPrompt = source.match(/default:I=([A-Za-z_$][\w$]*)\(\);break/);
+const defaultPrompt = source.match(/default:[A-Za-z_$][\w$]*=([A-Za-z_$][\w$]*)\(\);break/);
 if (defaultPrompt) basePromptNameBySymbol.set(defaultPrompt[1], 'smart');
 
 // Evaluate each candidate by extracting its source slice and running it in a fresh vm
@@ -285,7 +393,7 @@ console.log(`Rendered ${rendered.length} prompts.`);
 const MIN_BODY_CHARS = 200;
 const substantial = rendered.filter((r) => {
   const nonEmptyLines = r.text.split('\n').filter((l) => l.trim().length > 0);
-  return r.text.length >= MIN_BODY_CHARS && nonEmptyLines.length >= 2;
+  return r.text.length >= MIN_BODY_CHARS && r.text.length <= MAX_PROMPT_CHARS && nonEmptyLines.length >= 2 && !isBundleLikeText(r.text);
 });
 console.log(`Dropped ${rendered.length - substantial.length} trivial prompts.`);
 
@@ -462,7 +570,7 @@ named.forEach((block) => {
   const body = [
     `# ${block.derivedName}`,
     '',
-    `_Source: \`dist/main.js:${block.line}\` (symbol \`${block.name}\`)_`,
+    `_Source: \`${sourceLabel}:${block.line}\` (symbol \`${block.name}\`)_`,
     '',
     block.text,
     '',
@@ -489,7 +597,7 @@ function replaceGeneratedCatalog(catalog) {
 }
 
 const catalog = [
-  `Source: ${path.relative(ROOT, BUNDLE)}`,
+  `Source: ${sourceLabel}`,
   `Package: ${pkg.name}@${pkg.version}`,
   '',
   'Notes:',

package.json

@@ -13,6 +13,6 @@
   "dependencies": {
     "@babel/parser": "^7.29.3",
     "@babel/types": "^7.29.0",
-    "@sourcegraph/amp": "0.0.1781003299-g2fde29"
+    "@ampcode/cli": "0.0.1781102632-gaaab69"
   }
 }