docdex 0.2.74 → 0.2.76

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -1,5 +1,18 @@
1
1
  # Changelog
2
2
 
3
+ ## Unreleased
4
+
5
+ ## 0.2.76
6
+ - Add packaged provider-neutral local LLM setup and diagnostics guidance so npm installs do not imply Ollama is mandatory when another supported local service or mcoda agent already exists.
7
+ - Add packaged migration/test coverage notes for preserving explicit user model/provider/agent settings while replacing generated legacy Ollama defaults.
8
+ - Bump packaged release metadata to 0.2.76.
9
+
10
+ ## 0.2.75
11
+ - Add packaged automatic AI-terminal capture, generated-skill registry, validation, sync, install, rollback, and context-hint surfaces for personal-preferences and mind-clone workflows.
12
+ - Add packaged generated-skill quality reports with replay, drift, stale-source, activation feedback, and promote/keep/review/demote/quarantine recommendations.
13
+ - Split packaged personal-preferences storage, digestion, routines, clone, governance, operator-event, and generated-skill logic into focused modules with HTTP, CLI, MCP, and test coverage.
14
+ - Bump packaged release metadata to 0.2.75.
15
+
3
16
  ## 0.2.74
4
17
  - Add packaged compatibility HTTP aliases for repo memory, profile memory, personal-preferences, conversation, diary, KG, index, tree, files, stats, repo-inspect, impact diagnostics, and retrieval-only RAG surfaces.
5
18
  - Add packaged delete endpoints for repo memory, profile preferences, and diary entries, including storage cleanup and HTTP regression coverage.
package/README.md CHANGED
@@ -34,7 +34,7 @@ Most AI tools rely on "grep" (fast but dumb) or hosted RAG (slow and requires up
34
34
  * **💾 Repo Memory:** Stores project facts, decisions, and notes locally.
35
35
  * **👤 Agent Memory:** Remembers user preferences (e.g., "Use concise bullet points") across different repositories.
36
36
  * **🔌 MCP Native:** Auto-configures for tools like Claude Desktop, Cursor, and Windsurf.
37
- * **🌐 Web Enrichment:** Optional web search with local LLM filtering (via Ollama).
37
+ * **🌐 Web Enrichment:** Optional web search with local LLM filtering through detected local LLM services.
38
38
 
39
39
  ---
40
40
 
@@ -212,13 +212,14 @@ docdexd profile add --agent-id "default" --category style --content "Use concise
212
212
 
213
213
  ```
214
214
 
215
- ### 3. Local LLM (Ollama)
215
+ ### 3. Local LLM Services
216
216
 
217
- Docdex uses Ollama for embeddings and optional local chat.
217
+ Docdex detects supported local LLM services before it suggests installing anything. It can reuse Ollama, vLLM, llama.cpp-compatible OpenAI endpoints, LM Studio, LocalAI, SGLang, TGI-compatible deployments, and healthy local mcoda agents when they are already present. Ollama remains the recommended fallback because it is the easiest guided setup path.
218
218
 
219
- * **Setup:** Run `docdex setup` for an interactive wizard.
220
- * **Manual:** Ensure `nomic-embed-text` is pulled in Ollama (`ollama pull nomic-embed-text`).
221
- * **Custom URL:**
219
+ * **Setup:** Run `docdex setup` for an interactive wizard that lists detected services, models, embedding candidates, and local delegation agents.
220
+ * **Inspect:** Run `docdexd llm detect --json` or `docdexd llm diagnostics --json` to see why a service/model was selected, skipped, or marked unhealthy.
221
+ * **Manual Ollama fallback:** If no usable service is installed, pull the fallback embedding model with `ollama pull nomic-embed-text`.
222
+ * **Custom Ollama URL:**
222
223
  ```bash
223
224
  DOCDEX_OLLAMA_BASE_URL=http://127.0.0.1:11434 docdexd daemon --host 127.0.0.1 --port 28491
224
225
 
package/assets/agents.md CHANGED
@@ -1,4 +1,4 @@
1
- ---- START OF DOCDEX INFO V0.2.74 ----
1
+ ---- START OF DOCDEX INFO V0.2.76 ----
2
2
  Docdex URL: http://127.0.0.1:28491
3
3
  Use this base URL for Docdex HTTP endpoints.
4
4
  Health check endpoint: `GET /healthz` (not `/v1/health`).
@@ -99,6 +99,7 @@ When unsure which memory lane fits the task, call `docdex_memory_route` first.
99
99
  - Use `docdex_memory_layers` when you need the full six-layer map and storage/scope details.
100
100
  - Repo memory and profile memory are the core/default lanes.
101
101
  - Conversation memory, diary memory, temporal knowledge graph, and personal-preferences memory are selective lanes for continuity, handoff notes, structured recall, and richer user-specific context.
102
+ - For non-trivial agent work where personal-preferences memory is relevant, call `docdex_clone_directive` before planning and compare completed work against its required steps and approval gates before the final response.
102
103
 
103
104
  ### D. Conversation Memory + Temporal Knowledge Graph
104
105
 
package/lib/install.js CHANGED
@@ -2433,7 +2433,7 @@ function printPostInstallBanner() {
2433
2433
  const plain = [
2434
2434
  "[docdex] Docdex installed successfully.",
2435
2435
  "[docdex] Next step: run `docdex setup` to complete the installation.",
2436
- "[docdex] Setup configures Ollama/models + browser."
2436
+ "[docdex] Setup detects local LLM services, offers the Ollama fallback, and configures browser."
2437
2437
  ].join("\r\n") + "\r\n";
2438
2438
  if (!writeDirect(plain)) {
2439
2439
  writeStderr(plain);
@@ -2449,7 +2449,7 @@ function printPostInstallBanner() {
2449
2449
  "",
2450
2450
  "\x1b[32mDocdex installed successfully!\x1b[0m",
2451
2451
  "\x1b[41m\x1b[97m IMPORTANT \x1b[0m \x1b[33mNext step:\x1b[0m run \x1b[32m`docdex setup`\x1b[0m to complete the installation.",
2452
- "\x1b[33mSetup:\x1b[0m configures Ollama/models + browser.",
2452
+ "\x1b[33mSetup:\x1b[0m detects local LLM services, offers the Ollama fallback, and configures browser.",
2453
2453
  "\x1b[34mTip:\x1b[0m after setup, the daemon should auto-start; if not, run \x1b[36m`docdexd daemon`\x1b[0m"
2454
2454
  ];
2455
2455
  if (process.platform === "win32") {
@@ -3057,7 +3057,13 @@ async function runPostInstallSetup({ binaryPath, logger, env, skipDaemon, distBa
3057
3057
  if (!setupLaunch.ok) {
3058
3058
  if (setupLaunch.reason !== "skipped") {
3059
3059
  log.warn?.("[docdex] setup wizard did not launch. Run `docdex setup`.");
3060
- recordSetupPending({ reason: setupLaunch.reason, port, repoRoot: daemonRoot });
3060
+ log.warn?.("[docdex] `docdex setup` will detect existing local LLM services before offering the Ollama fallback.");
3061
+ recordSetupPending({
3062
+ reason: setupLaunch.reason,
3063
+ port,
3064
+ repoRoot: daemonRoot,
3065
+ setup: "local_llm_service_choice"
3066
+ });
3061
3067
  }
3062
3068
  }
3063
3069
  return { port, url, configPath };
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "docdex",
3
- "version": "0.2.74",
3
+ "version": "0.2.76",
4
4
  "mcpName": "io.github.bekirdag/docdex",
5
5
  "description": "Local-first documentation and code indexer with HTTP/MCP search, AST, and agent memory.",
6
6
  "bin": {