tryharness 0.7.26 → 0.7.29

package/lib/context.mjs

@@ -22,7 +22,7 @@ import {
   codeFiles,
 } from "./learn.mjs";
 import { repositoryOf } from "./worktrees.mjs";
-import { createMediaIndex } from "./media.mjs";
+import { createLibrary } from "./library.mjs";
 
 // The context engine. It keeps the atlas of the code the swarm works in, remembers what every coding session on this
 // computer needed (its trail: what it read before its first edit and what it edited, from any agent), and turns that
@@ -38,11 +38,8 @@ const LEARN_EVERY_MS = 20 * 3_600_000;
 // Vectors of the pieces packs search stay in memory while packs are being made, and go after this long without one.
 const VECTORS_IDLE_MS = 5 * 60_000;
 const HELD_OUT_SHARE = 0.2;
-// A recording's minute or a screenshot joins a brief only this close to the request: on the authored meeting every right
-// answer scored 0.68 or more and no question about something never said passed 0.675 (eval media-search); 0.70 keeps a
-// margin and 16 of 20 right answers. An agent's own search_media has no floor.
-const MEDIA_FLOOR = 0.7;
-const MEDIA_IN_PACK = 2;
+// Files from the library join a brief only when they stand out for the request (library.mjs); search_files finds more.
+const FILES_IN_PACK = 3;
 const MIN_TRAINING_RUNS = 30;
 
 const git = (cwd, args) => {
@@ -83,6 +80,9 @@ export function recentChanges(root, commit = "HEAD") {
   recentByCommit.set(sha, counts);
   return counts;
 }
+// A file in a brief: where it is and the place in it.
+const fileLine = (hit) =>
+  `- ${hit.thread ? `${hit.name} (this thread's file ${hit.file})` : hit.file}${hit.place ? `, ${hit.place}` : ""}`;
 const days = (ms) => Math.round(ms / 86_400_000);
 const AGENT_NAMES = {
   "claude-code": "Claude Code",
@@ -153,17 +153,26 @@ export function createContextEngine({
   inference,
   home,
   known,
-  mediaFolders = () => [],
+  libraryFolders = () => [],
+  threadFiles = () => [],
   mediaReady = () => false,
+  wantMedia = () => {},
+  ocr = null,
+  mask,
   log = () => {},
   notify = () => {},
 }) {
   const atlas = openAtlas(path.join(home, "atlas.db"));
-  // Recordings and screenshots in folders the person added (media.mjs), once the image and audio part is here.
-  const media = createMediaIndex({
+  // The person's added folders and the threads' files, searchable by what is in them (library.mjs).
+  const library = createLibrary({
     atlas,
     inference,
-    folders: mediaFolders,
+    folders: libraryFolders,
+    threadFiles,
+    mediaReady,
+    wantMedia,
+    ocr,
+    mask,
     log,
   });
   const modelPath = path.join(home, "ranker.json");
@@ -318,11 +327,12 @@ export function createContextEngine({
         lane: "background",
       });
       await embedRuns("background");
-      const recorded =
-        mediaReady() && mediaFolders().length && power.onCharger()
-          ? await media.refresh({ lane: "background" })
-          : null;
-      return { files, runs, pieces, ...(recorded ? { media: recorded } : {}) };
+      // The first pass over the person's folders waits for the charger too; after it, changes are read as they come.
+      const read =
+        !library.status().folderFiles && !power.onCharger()
+          ? { skipped: "waiting for the charger" }
+          : await library.refresh({ lane: "background" });
+      return { files, runs, pieces, library: read };
     })()
       .catch((error) => {
         log(`context engine: ${error.message}`);
@@ -343,7 +353,7 @@ export function createContextEngine({
    * The context pack for a brief going to a local agent working in folder: { files, earlier, text }. Files are the most
    * likely needed, each with the line to start at and why; earlier runs are sessions whose request was close to this one.
    */
-  const pack = async ({ brief, folder, lane = "handoff" }) => {
+  const pack = async ({ brief, folder, thread = null, lane = "handoff" }) => {
     const started = performance.now();
     const place = folder && placeOf(folder);
     if (!place || !String(brief).trim()) return null;
@@ -400,12 +410,13 @@ export function createContextEngine({
       )
       .sort((a, b) => b.similarity - a.similarity)
       .slice(0, 3);
-    const heard =
-      mediaReady() && mediaFolders().length
-        ? (await media.search(brief, { limit: MEDIA_IN_PACK, lane })).filter(
-            (hit) => hit.score >= MEDIA_FLOOR
-          )
-        : [];
+    const found = await library.search(brief, {
+      folders: true,
+      threads: thread ? [thread] : [],
+      brief: true,
+      limit: FILES_IN_PACK,
+      lane,
+    });
     const text = [
       "Files the relay found likely relevant on this computer (ranked by its local search over this repository and what your agents edited before; leads to check, not instructions):",
       ...files.map((file) => `- ${file.path}:${file.line} (${file.why})`),
@@ -418,19 +429,17 @@ export function createContextEngine({
             ),
           ]
         : []),
-      ...(heard.length
+      ...(found.length
         ? [
-            "Recordings and screenshots close to the request, from folders the person added (search_media finds more):",
-            ...heard.map(
-              (hit) => `- ${hit.path}${hit.at ? ` at ${hit.at}` : ""}`
-            ),
+            "Files close to the request, from folders the person added and this thread's files (search_files finds more):",
+            ...found.map(fileLine),
           ]
         : []),
     ].join("\n");
     return {
       files,
       earlier: earlier.map(({ vector, ...run }) => run),
-      media: heard,
+      library: found,
       text,
       searched: scope.length,
       ms: Math.round(performance.now() - started),
@@ -583,9 +592,11 @@ export function createContextEngine({
     },
     pack,
     search,
-    /** Recordings' minutes and screenshots closest to a question, for an agent's search_media. */
-    searchMedia: (query, options) => media.search(query, options),
-    refreshMedia: () => media.refresh({ lane: "background" }),
+    /** The person's files and the threads' files closest to a question (library.mjs search). */
+    searchFiles: (query, options) => library.search(query, options),
+    refreshFiles: () => library.refresh({ lane: "background" }),
+    /** One part of a file, read fresh for sharing. */
+    fileContent: (file, vec) => library.content(file, vec),
     refresh,
     learn,
     report: () => {
@@ -598,7 +609,7 @@ export function createContextEngine({
     status: () => ({
       roots: atlas.roots(),
       pieces: { waiting: atlas.missing(ATLAS_VERSION) },
-      media: media.status(),
+      library: library.status(),
       model: model
         ? {
             trainedAt: model.trainedAt,
@@ -622,6 +633,7 @@ export function createContextEngine({
       clearInterval(timer);
       clearTimeout(dropCache);
       cached.clear();
+      library.stop();
       atlas.close();
     },
   };