fix(doctor): resolve false positive for local memory search when no explicit modelPath by adhishthite · Pull Request #32014 · openclaw/openclaw

adhishthite · 2026-03-02T18:01:56Z

Summary

Fixes a false positive warning in openclaw doctor where it reports "Memory search provider is set to "local" but no local model file was found" even when local memory search is working correctly with the default auto-resolved model.

Problem

When memorySearch.provider is "local" (or "auto") and no explicit local.modelPath is configured, the runtime in createLocalEmbeddingProvider() auto-resolves to DEFAULT_LOCAL_MODEL (hf:ggml-org/embeddinggemma-300m-qat-q8_0-GGUF). However, the doctor's hasLocalEmbeddings() only inspected the config value and returned false when modelPath was empty, triggering a misleading warning.

openclaw memory status --deep confirms everything is working:

Provider: local (requested: local)
Model: hf:ggml-org/embeddinggemma-300m-qat-q8_0-GGUF/embeddinggemma-300m-qat-Q8_0.gguf
Embeddings: ready
48/48 files indexed, 212 chunks

Fix

In hasLocalEmbeddings(), fall back to DEFAULT_LOCAL_MODEL when no explicit modelPath is set — matching the runtime behavior in createLocalEmbeddingProvider():

// Before
const modelPath = local.modelPath?.trim();

// After
const modelPath = local.modelPath?.trim() || DEFAULT_LOCAL_MODEL;

Tests

Added test: local provider with no explicit modelPath should not warn (default model fallback)
Added test: local provider with explicit hf: modelPath should not warn
Updated auto-mode test to reflect that default local model is now correctly detected
All 11 tests pass

Closes #31998

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3cbe22737e

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

src/commands/doctor-memory-search.ts

greptile-apps · 2026-03-02T18:05:46Z

Greptile Summary

Fixes a false positive in openclaw doctor where it incorrectly reported missing local model files even when the default auto-resolved model was available. The fix ensures the doctor's hasLocalEmbeddings() check matches the runtime behavior in createLocalEmbeddingProvider() by falling back to DEFAULT_LOCAL_MODEL when no explicit modelPath is configured.

Correctly mirrors runtime logic that auto-resolves to the default HuggingFace model (hf:ggml-org/embeddinggemma-300m-qat-q8_0-GGUF)
Added comprehensive test coverage for both no-modelPath and explicit hf: modelPath scenarios
Updated auto-mode test to reflect that default local model is now properly detected

Confidence Score: 5/5

This PR is safe to merge with minimal risk
The fix is a simple one-line change that adds proper fallback logic to match existing runtime behavior. It's well-tested with three test cases covering the scenarios, and the change is isolated to the doctor health check functionality without affecting runtime behavior
No files require special attention

_{Last reviewed commit: 3cbe227}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 43ca93b385

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-03-02T18:09:43Z

src/commands/doctor-memory-search.ts

+      if (hasLocalEmbeddings(resolved.local, true)) {
+        return; // local model file exists (or default model will be auto-downloaded)


Preserve probe-based warning for local default model path

This new early return bypasses gatewayMemoryProbe handling whenever provider is "local" and modelPath is unset, because hasLocalEmbeddings(..., true) now treats DEFAULT_LOCAL_MODEL as immediately available. In doctor.ts, this function is called with probe data, so when the gateway reports ready: false (for example, local embeddings cannot initialize due to missing node-llama-cpp or model download/setup failure), doctor now emits no memory-search warning and can report a broken local setup as healthy.

Useful? React with 👍 / 👎.

adhishthite · 2026-03-02T18:32:52Z

CI Failures — Pre-existing, Unrelated to This PR

The two test failures in this PR are not caused by our changes. Both fail on main as well:

1. `isolated-agent.uses-last-non-empty-agent-text-as.test.ts` (line 424)

AssertionError: expected 'anthropic' to be 'openai'

The test expects openai as the default provider after a model override round-trip, but the model resolution logic now returns anthropic. Likely a recent change to the model catalog or default provider resolution.

2. `temp-path-guard.test.ts` (line 239)

Expected: []
Received: ["src/gateway/server.auth.control-ui.suite.ts",
           "src/gateway/server.auth.default-token.suite.ts", 
           "src/gateway/server.auth.shared.ts"]

Three auth test files use dynamic tmpdir joins that trigger the security guardrail scanner. These files likely need to use the repo's safe tmpdir helper.

Our change (doctor-memory-search.ts + tests) passes in all runners. The check job (lint + type-check) is green ✅.

…xplicit modelPath When memorySearch.provider is 'local' (or 'auto') and no explicit local.modelPath is configured, the runtime auto-resolves to DEFAULT_LOCAL_MODEL (embeddinggemma-300m via HuggingFace). However, the doctor's hasLocalEmbeddings() check only inspected the config value and returned false when modelPath was empty, triggering a misleading warning. Fix: fall back to DEFAULT_LOCAL_MODEL in hasLocalEmbeddings(), matching the runtime behavior in createLocalEmbeddingProvider(). Closes openclaw#31998

Address review feedback: canAutoSelectLocal() in the runtime skips local for empty/hf: model paths in auto mode. The DEFAULT_LOCAL_MODEL fallback should only apply when provider is explicitly 'local', not when provider is 'auto' — otherwise users with no local file and no API keys would get a clean doctor report but no working embeddings. Add useDefaultFallback parameter to hasLocalEmbeddings() to distinguish the two code paths.

…odel When hasLocalEmbeddings returns true via DEFAULT_LOCAL_MODEL fallback, also check the gateway memory probe if available. If the probe reports not-ready (e.g. node-llama-cpp missing or model download failed), emit a warning instead of silently reporting healthy. Addresses review feedback about bypassing probe-based validation.

@adhishthite

…w#32014) (thanks @adhishthite)

steipete · 2026-03-02T18:35:43Z

Landed via temp rebase onto main.

Gate: bunx vitest run src/commands/doctor-memory-search.test.ts
Land commit: b6df323
Merge commit: 63734df

Thanks @adhishthite!

@adhishthite

…xplicit modelPath (openclaw#32014) * fix(doctor): resolve false positive for local memory search when no explicit modelPath When memorySearch.provider is 'local' (or 'auto') and no explicit local.modelPath is configured, the runtime auto-resolves to DEFAULT_LOCAL_MODEL (embeddinggemma-300m via HuggingFace). However, the doctor's hasLocalEmbeddings() check only inspected the config value and returned false when modelPath was empty, triggering a misleading warning. Fix: fall back to DEFAULT_LOCAL_MODEL in hasLocalEmbeddings(), matching the runtime behavior in createLocalEmbeddingProvider(). Closes openclaw#31998 * fix: scope DEFAULT_LOCAL_MODEL fallback to explicit provider:local only Address review feedback: canAutoSelectLocal() in the runtime skips local for empty/hf: model paths in auto mode. The DEFAULT_LOCAL_MODEL fallback should only apply when provider is explicitly 'local', not when provider is 'auto' — otherwise users with no local file and no API keys would get a clean doctor report but no working embeddings. Add useDefaultFallback parameter to hasLocalEmbeddings() to distinguish the two code paths. * fix: preserve gateway probe warning for local provider with default model When hasLocalEmbeddings returns true via DEFAULT_LOCAL_MODEL fallback, also check the gateway memory probe if available. If the probe reports not-ready (e.g. node-llama-cpp missing or model download failed), emit a warning instead of silently reporting healthy. Addresses review feedback about bypassing probe-based validation. * fix: add changelog attribution for doctor local fallback fix (openclaw#32014) (thanks @adhishthite) --------- Co-authored-by: Adhish <adhishthite@Adhishs-MacBook-Pro.local> Co-authored-by: Peter Steinberger <steipete@gmail.com>

@adhishthite

…xplicit modelPath (openclaw#32014) * fix(doctor): resolve false positive for local memory search when no explicit modelPath When memorySearch.provider is 'local' (or 'auto') and no explicit local.modelPath is configured, the runtime auto-resolves to DEFAULT_LOCAL_MODEL (embeddinggemma-300m via HuggingFace). However, the doctor's hasLocalEmbeddings() check only inspected the config value and returned false when modelPath was empty, triggering a misleading warning. Fix: fall back to DEFAULT_LOCAL_MODEL in hasLocalEmbeddings(), matching the runtime behavior in createLocalEmbeddingProvider(). Closes openclaw#31998 * fix: scope DEFAULT_LOCAL_MODEL fallback to explicit provider:local only Address review feedback: canAutoSelectLocal() in the runtime skips local for empty/hf: model paths in auto mode. The DEFAULT_LOCAL_MODEL fallback should only apply when provider is explicitly 'local', not when provider is 'auto' — otherwise users with no local file and no API keys would get a clean doctor report but no working embeddings. Add useDefaultFallback parameter to hasLocalEmbeddings() to distinguish the two code paths. * fix: preserve gateway probe warning for local provider with default model When hasLocalEmbeddings returns true via DEFAULT_LOCAL_MODEL fallback, also check the gateway memory probe if available. If the probe reports not-ready (e.g. node-llama-cpp missing or model download failed), emit a warning instead of silently reporting healthy. Addresses review feedback about bypassing probe-based validation. * fix: add changelog attribution for doctor local fallback fix (openclaw#32014) (thanks @adhishthite) --------- Co-authored-by: Adhish <adhishthite@Adhishs-MacBook-Pro.local> Co-authored-by: Peter Steinberger <steipete@gmail.com>

@adhishthite

…xplicit modelPath (openclaw#32014) * fix(doctor): resolve false positive for local memory search when no explicit modelPath When memorySearch.provider is 'local' (or 'auto') and no explicit local.modelPath is configured, the runtime auto-resolves to DEFAULT_LOCAL_MODEL (embeddinggemma-300m via HuggingFace). However, the doctor's hasLocalEmbeddings() check only inspected the config value and returned false when modelPath was empty, triggering a misleading warning. Fix: fall back to DEFAULT_LOCAL_MODEL in hasLocalEmbeddings(), matching the runtime behavior in createLocalEmbeddingProvider(). Closes openclaw#31998 * fix: scope DEFAULT_LOCAL_MODEL fallback to explicit provider:local only Address review feedback: canAutoSelectLocal() in the runtime skips local for empty/hf: model paths in auto mode. The DEFAULT_LOCAL_MODEL fallback should only apply when provider is explicitly 'local', not when provider is 'auto' — otherwise users with no local file and no API keys would get a clean doctor report but no working embeddings. Add useDefaultFallback parameter to hasLocalEmbeddings() to distinguish the two code paths. * fix: preserve gateway probe warning for local provider with default model When hasLocalEmbeddings returns true via DEFAULT_LOCAL_MODEL fallback, also check the gateway memory probe if available. If the probe reports not-ready (e.g. node-llama-cpp missing or model download failed), emit a warning instead of silently reporting healthy. Addresses review feedback about bypassing probe-based validation. * fix: add changelog attribution for doctor local fallback fix (openclaw#32014) (thanks @adhishthite) --------- Co-authored-by: Adhish <adhishthite@Adhishs-MacBook-Pro.local> Co-authored-by: Peter Steinberger <steipete@gmail.com>

@adhishthite

…xplicit modelPath (openclaw#32014) * fix(doctor): resolve false positive for local memory search when no explicit modelPath When memorySearch.provider is 'local' (or 'auto') and no explicit local.modelPath is configured, the runtime auto-resolves to DEFAULT_LOCAL_MODEL (embeddinggemma-300m via HuggingFace). However, the doctor's hasLocalEmbeddings() check only inspected the config value and returned false when modelPath was empty, triggering a misleading warning. Fix: fall back to DEFAULT_LOCAL_MODEL in hasLocalEmbeddings(), matching the runtime behavior in createLocalEmbeddingProvider(). Closes openclaw#31998 * fix: scope DEFAULT_LOCAL_MODEL fallback to explicit provider:local only Address review feedback: canAutoSelectLocal() in the runtime skips local for empty/hf: model paths in auto mode. The DEFAULT_LOCAL_MODEL fallback should only apply when provider is explicitly 'local', not when provider is 'auto' — otherwise users with no local file and no API keys would get a clean doctor report but no working embeddings. Add useDefaultFallback parameter to hasLocalEmbeddings() to distinguish the two code paths. * fix: preserve gateway probe warning for local provider with default model When hasLocalEmbeddings returns true via DEFAULT_LOCAL_MODEL fallback, also check the gateway memory probe if available. If the probe reports not-ready (e.g. node-llama-cpp missing or model download failed), emit a warning instead of silently reporting healthy. Addresses review feedback about bypassing probe-based validation. * fix: add changelog attribution for doctor local fallback fix (openclaw#32014) (thanks @adhishthite) --------- Co-authored-by: Adhish <adhishthite@Adhishs-MacBook-Pro.local> Co-authored-by: Peter Steinberger <steipete@gmail.com>

openclaw-barnacle bot added commands Command implementations size: XS labels Mar 2, 2026

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

src/commands/doctor-memory-search.ts Outdated Show resolved Hide resolved

openclaw-barnacle bot added size: S and removed size: XS labels Mar 2, 2026

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

adhishthite mentioned this pull request Mar 2, 2026

Doctor: avoid false local memory warning without modelPath #32020

Closed

Adhish and others added 4 commits March 2, 2026 18:34

fix: add changelog attribution for doctor local fallback fix (opencla…

b6df323

…w#32014) (thanks @adhishthite)

steipete force-pushed the fix/doctor-local-memory-search-false-positive branch from cd7510a to b6df323 Compare March 2, 2026 18:35

steipete merged commit 63734df into openclaw:main Mar 2, 2026
9 checks passed

steipete mentioned this pull request Mar 2, 2026

fix(doctor): avoid false warning for local memory defaults #32042

Closed

18 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(doctor): resolve false positive for local memory search when no explicit modelPath#32014

fix(doctor): resolve false positive for local memory search when no explicit modelPath#32014
steipete merged 4 commits intoopenclaw:mainfrom
adhishthite:fix/doctor-local-memory-search-false-positive

adhishthite commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

Uh oh!

greptile-apps bot commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Uh oh!

adhishthite commented Mar 2, 2026

Uh oh!

Uh oh!

steipete commented Mar 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		if (hasLocalEmbeddings(resolved.local, true)) {
		return; // local model file exists (or default model will be auto-downloaded)

Uh oh!

Conversation

adhishthite commented Mar 2, 2026

Summary

Problem

Fix

Tests

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

greptile-apps bot commented Mar 2, 2026

Greptile Summary

Confidence Score: 5/5

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Choose a reason for hiding this comment

Uh oh!

adhishthite commented Mar 2, 2026

CI Failures — Pre-existing, Unrelated to This PR

1. isolated-agent.uses-last-non-empty-agent-text-as.test.ts (line 424)

2. temp-path-guard.test.ts (line 239)

Uh oh!

Uh oh!

steipete commented Mar 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

1. `isolated-agent.uses-last-non-empty-agent-text-as.test.ts` (line 424)

2. `temp-path-guard.test.ts` (line 239)