Skip to content

Conversation

@rabi
Copy link
Contributor

@rabi rabi commented Jan 15, 2026

Summary

The check that disabled subagents for Gemini models seems to have been added as a workaround for thought_signature issues. Now that the underlying issues have been resolved, remove the check to restore subagent functionality for Gemini providers.

Type of Change

  • Feature
  • Bug fix
  • Refactor / Code quality
  • Performance improvement
  • Documentation
  • Tests
  • Security fix
  • Build / Release
  • Other (specify below)

Testing

Tested with gemini models.

starting session | provider: google model: gemini-3-pro-preview
    session id: 20260115_85
    working directory: /tmp/goose

goose is running! Enter your instructions, or try asking what goose can do.

Context: ○○○○○○○○○○ 0% (0/128000 tokens)
( O)> Use subagent with google provider and gemini-3-pro-preview model to fetch last 10 PRs from vllm-project/vllm

─── subagent |  ──────────────────────────
instructions: Fetch the last 10 pull requests from the repository 'vllm-project/vllm'. List the PR number, titl...
settings:
    model: gemini-3-pro-preview
    provider: google

Here are the last 10 pull requests from `vllm-project/vllm`:

| PR # | Title | Author | State |
|------|-------|--------|-------|
| 32403 | [Performance] Improve Triton prefill attention kernel's performance | Isotr0py | open |
| 32398 | [KVConnector] OffloadingConnector: Add preemptions-only mode | orozery | open |
| 32397 | [Model] Enable LoRA support for internvl2 | MatteoFari | open |
| 32396 | [Refactor] [11/N] to simplify the mcp architecture | chaunceyjiang | open |
| 32395 | [Frontend][1/n] Make pooling entrypoints request schema consensus. | noooop | open |
| 32394 | fix(tool_parsers): enable type conversion for Seed OSS tool parser streaming mode | karanb192 | open |
| 32393 | Support custom URI schemes and trace handlers for profiler | diviramon | open |
| 32392 | Support custom URI schemes and trace handlers for profiler | brad-mengchi | closed |
| 32390 | [Model] Add OpenVLA model support | PalmDr | open |
| 32389 | [Model] Avoid token selection in SigLIP pooling head | DarkLight1337 | closed |

@jamadeo jamadeo requested a review from DOsinga January 15, 2026 20:22
@jamadeo
Copy link
Collaborator

jamadeo commented Jan 15, 2026

paging @DOsinga as he probably saw the worst of the gemini compatibility with subagent tools. Was it just thought signature related, Douwe?

The check that disabled subagents for Gemini models was added as a
workaround for thought_signature issues. Now that the underlying
issue has been resolved, remove the check to restore subagent
functionality for Gemini providers.

Signed-off-by: rabi <[email protected]>
@rabi
Copy link
Contributor Author

rabi commented Jan 16, 2026

paging @DOsinga as he probably saw the worst of the gemini compatibility with subagent tools. Was it just thought signature related, Douwe?

Thanks @jamadeo, I probably did some simple tests. Anything specific that was not working, I can check again.

@DOsinga
Copy link
Collaborator

DOsinga commented Jan 20, 2026

I think it was actually something else where sub-agents had some hand craften json spec that made gemini choke, so all good!

@DOsinga DOsinga merged commit 932923e into block:main Jan 28, 2026
18 checks passed
lifeizhou-ap added a commit that referenced this pull request Jan 28, 2026
* main: (47 commits)
  Upgrade error handling (#6747)
  Fix/filter audience 6703 local (#6773)
  chore: re-sync package-lock.json (#6783)
  upgrade electron to 39.3.0 (#6779)
  allow skipping providers in test_providers.sh (#6778)
  fix: enable custom model entry for OpenRouter provider (#6761)
  Remove codex skills flag support (#6775)
  Improve mcp test (#6671)
  Feat/anthropic custom headers (#6774)
  Fix/GitHub copilot error handling 5845 (#6771)
  fix(ui): respect width parameter in MCP app size-changed notifications (#6376)
  fix: address compilation issue in main (#6776)
  Upgrade GitHub Actions for Node 24 compatibility (#6699)
  fix(google): preserve thought signatures in streaming responses (#6708)
  added reduce motion support for css animations and streaming text (#6551)
  fix: Re-enable subagents for Gemini models (#6513)
  fix(google): use parametersJsonSchema for full JSON Schema support (#6555)
  fix: respect GOOSE_CLI_MIN_PRIORITY for shell streaming output (#6558)
  feat: add requires_auth flag for custom providers without authentication (#6705)
  fix: normalize extension names consistently in ExtensionManager (#6529)
  ...
michaelneale added a commit that referenced this pull request Jan 29, 2026
* main: (30 commits)
  Different approach to determining final confidence level of prompt injection evaluation outcomes (#6729)
  fix: read_resource_tool deadlock causing test_compaction to hang (#6737)
  Upgrade error handling (#6747)
  Fix/filter audience 6703 local (#6773)
  chore: re-sync package-lock.json (#6783)
  upgrade electron to 39.3.0 (#6779)
  allow skipping providers in test_providers.sh (#6778)
  fix: enable custom model entry for OpenRouter provider (#6761)
  Remove codex skills flag support (#6775)
  Improve mcp test (#6671)
  Feat/anthropic custom headers (#6774)
  Fix/GitHub copilot error handling 5845 (#6771)
  fix(ui): respect width parameter in MCP app size-changed notifications (#6376)
  fix: address compilation issue in main (#6776)
  Upgrade GitHub Actions for Node 24 compatibility (#6699)
  fix(google): preserve thought signatures in streaming responses (#6708)
  added reduce motion support for css animations and streaming text (#6551)
  fix: Re-enable subagents for Gemini models (#6513)
  fix(google): use parametersJsonSchema for full JSON Schema support (#6555)
  fix: respect GOOSE_CLI_MIN_PRIORITY for shell streaming output (#6558)
  ...
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants