fix(agents): skip Ollama discovery when explicit models configured by lailoo · Pull Request #28793 · openclaw/openclaw

lailoo · 2026-02-27T14:33:17Z

Summary

Skip Ollama model discovery when explicit models are already configured in models.providers.ollama, preventing unnecessary 5s timeout delays at startup for remote Ollama hosts.

Problem

When users configure explicit models in models.providers.ollama (e.g. for a remote Ollama host), resolveImplicitProviders still calls discoverOllamaModels() which makes an HTTP request to the Ollama API with a 5s timeout. For remote hosts that may be unreachable at startup, this adds a guaranteed 5s delay. The docs state that explicit config should skip auto-discovery, but the code didn't honor this.

Changes

src/agents/models-config.providers.ts: Check if explicitProviders.ollama.models is non-empty before running discovery. When explicit models exist, use them directly instead of calling discoverOllamaModels().
src/agents/models-config.providers.ollama.test.ts: Add regression test verifying that explicit models are preserved and discovery is skipped.

Test plan

Added test: "should skip discovery when explicit models are configured" — passes explicit models via explicitProviders.ollama.models and asserts they are returned as-is without discovery.
All existing Ollama provider tests continue to pass.
pnpm format:check and pnpm lint pass.

Effect on User Experience

Users with explicit Ollama model configs (especially remote hosts) no longer experience a 5s startup delay from unnecessary model discovery. No behavior change for users relying on auto-discovery (no explicit models configured).

Closes #28762

greptile-apps · 2026-02-27T14:37:28Z

Greptile Summary

Correctly skips Ollama model discovery when explicit models are configured in models.providers.ollama, eliminating unnecessary 5-second timeout delays for remote Ollama hosts.

Key Changes:

Added conditional check in src/agents/models-config.providers.ts to detect when explicitProviders.ollama.models is a non-empty array
When explicit models exist, the provider config is used directly without calling discoverOllamaModels()
When no explicit models are configured, discovery runs as before (maintains backward compatibility)
Added comprehensive regression test to verify explicit models are preserved and discovery is skipped

Impact:

Users with explicit Ollama configurations (especially remote hosts) no longer experience startup delays
No behavior change for users relying on auto-discovery
Aligns implementation with documented behavior that explicit config should skip auto-discovery

Confidence Score: 5/5

This PR is safe to merge with no identified issues
The implementation is straightforward and correct: it checks for non-empty explicit models array before running discovery, uses explicit config when available, and falls back to discovery otherwise. The test properly validates the fix by confirming explicit models are preserved and discovery is skipped. The change maintains backward compatibility and only affects the specific scenario it targets (explicit models configured).
No files require special attention

_{Last reviewed commit: 0e4a046}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8d9fc29f1f

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 · 2026-02-27T14:39:15Z

+    if (hasExplicitModels) {
+      // Use explicit models as-is, skip discovery
+      providers.ollama = { ...explicitOllama, apiKey: ollamaKey };


Preserve Ollama API default when skipping discovery

When explicitProviders.ollama.models is non-empty, this branch copies the explicit provider as-is and no longer injects api: "ollama". Because api is optional in config, valid explicit configs can omit it; with this change those configs end up with no provider/model API, so downstream logic like run/attempt.ts (which only enables native Ollama transport when model.api === "ollama") stops using the Ollama-native path. Before this commit, buildOllamaProvider() supplied the default API even for explicit-model merges.

Useful? React with 👍 / 👎.

Kansodata · 2026-02-27T14:57:55Z

I reproduced the failing check lane and prepared a minimal fix for the TypeScript typing error in the new Ollama test (input: string[] vs (\

Kansodata · 2026-02-27T14:58:08Z

Correction / clean summary:

I reproduced the failing check lane and prepared a minimal fix for the TypeScript typing error in the new Ollama test (input: string[] vs ("text"|"image")[]).

Support PR:

fix(agents): skip Ollama discovery when explicit models are configured #28827

If helpful, I can also fold this into your branch directly.

lailoo · 2026-02-27T15:10:22Z

@Kansodata Thanks for catching the typing issue and preparing the fix! That would be great — please go ahead and fold it into my branch directly. Appreciate the help 🙏

nikolasdehor

Correct optimization — skipping unnecessary Ollama discovery when explicit models are configured eliminates the 5s timeout. Type fix from #28827 is needed before merge.

…penclaw#28762)

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fc859b223c

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 · 2026-02-27T18:30:24Z

+    if (hasExplicitModels) {
+      // Use explicit models as-is, skip discovery
+      providers.ollama = { api: "ollama", ...explicitOllama, apiKey: ollamaKey };
+    } else {


Preserve default Ollama base URL when skipping discovery

When explicit Ollama models are present, this branch now copies explicitOllama directly and skips buildOllamaProvider(), so configs that intentionally omit models.providers.ollama.baseUrl no longer inherit the previous default http://127.0.0.1:11434. That changes runtime metadata for those models (their baseUrl becomes unset), and downstream local-model logic like isLocalBaseUrl() in src/commands/models/shared.ts will classify them as non-local, causing openclaw models list --local filters to drop Ollama entries that previously appeared.

Useful? React with 👍 / 👎.

vincentkoc · 2026-02-28T17:09:07Z

Thanks for the implementation here.

I consolidated the active merge path into #28827 (rebased on current main) so we have a single PR to land for #28762. Your work in this PR is preserved as the foundational implementation direction, and credit is called out explicitly in #28827.

I’m closing this PR as superseded by #28827. If there is any behavior gap, say the word and we can reopen/re-check quickly.

openclaw-barnacle Bot added agents Agent runtime and tooling size: S experienced-contributor labels Feb 27, 2026

chatgpt-codex-connector Bot reviewed Feb 27, 2026

View reviewed changes

Kansodata mentioned this pull request Feb 27, 2026

fix(agents): skip Ollama discovery when explicit models are configured #28827

Merged

nikolasdehor approved these changes Feb 27, 2026

View reviewed changes

fix(agents): skip Ollama discovery when explicit models configured (o…

fc859b2

…penclaw#28762)

chatgpt-codex-connector Bot reviewed Feb 27, 2026

View reviewed changes

chore: fix CONTRIBUTING.md formatting (pre-existing)

b293da6

vincentkoc closed this Feb 28, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(agents): skip Ollama discovery when explicit models configured#28793

fix(agents): skip Ollama discovery when explicit models configured#28793
lailoo wants to merge 2 commits intoopenclaw:mainfrom
lailoo:fix/ollama-discovery-timeout-28762

lailoo commented Feb 27, 2026

Uh oh!

greptile-apps Bot commented Feb 27, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Feb 27, 2026

Uh oh!

Kansodata commented Feb 27, 2026

Uh oh!

Kansodata commented Feb 27, 2026

Uh oh!

lailoo commented Feb 27, 2026

Uh oh!

nikolasdehor left a comment

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Feb 27, 2026

Uh oh!

vincentkoc commented Feb 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

lailoo commented Feb 27, 2026

Summary

Problem

Changes

Test plan

Effect on User Experience

Uh oh!

greptile-apps Bot commented Feb 27, 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 Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Kansodata commented Feb 27, 2026

Uh oh!

Kansodata commented Feb 27, 2026

Uh oh!

lailoo commented Feb 27, 2026

Uh oh!

nikolasdehor left a comment

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

vincentkoc commented Feb 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants