Fixes#291
The gemini CLI configuration had a duplicate -o json flag that caused
Gemini to return plain text instead of JSON, breaking the clink parser.
Root cause:
- Internal defaults (clink/constants.py) already provide -o json
- Config file (conf/cli_clients/gemini.json) duplicated this flag
- Result: command became 'gemini -o json ... -o json' (duplicate)
This caused Gemini CLI to malfunction and output plain text, leading to
'Failed to decode Gemini CLI JSON output' errors.
Solution:
Remove the duplicate -o json from additional_args since it's already
provided by internal defaults, matching the pattern used by claude and
codex CLI configurations.
After fix:
- Command: gemini -o json --telemetry false --yolo (correct)
- Output: Valid JSON with response and stats
- Parser: Successfully extracts content and metadata
The Claude CLI recently changed its configuration commands from
'claude config add-server' to 'claude mcp add -s user' and from
'claude config list' to 'claude mcp list'. This updates the
PowerShell setup script to detect and configure the zen server
using the new command syntax, fixing configuration detection
for Windows users.
Changes:
- Updated Test-ClaudeCliIntegration to use 'claude mcp list'
- Updated instructions to use 'claude mcp add -s user' syntax
The gemini CLI was returning plain text instead of JSON, causing clink to fail with:
'Failed to decode Gemini CLI JSON output: Expecting value: line 1 column 1 (char 0)'
Root cause: gemini.json was missing the '-o json' flag in additional_args.
This fix adds '-o json' to ensure JSON output format, aligning with the codex pattern (which uses --json).
Tested with gemini CLI v0.8.2
Stay in codex, plan review and fix complicated bugs, then ask it to spawn claude code and implement the plan.
This uses your current subscription instead of API tokens.
Stay in codex, plan review and fix complicated bugs, then ask it to spawn claude code and implement the plan.
This uses your current subscription instead of API tokens.
The previous commit (f98046c) added shutil.which() to resolve executables,
which broke two tests that only mocked subprocess execution. This commit
adds shutil.which() mocking to both test files to restore test compatibility.
Co-authored-by: Claude <noreply@anthropic.com>
model definitions now support a new `allow_code_generation` flag, only to be used with higher reasoning models such as GPT-5-Pro and-Gemini 2.5-Pro
When `true`, the `chat` tool can now request the external model to generate a full implementation / update / instructions etc and then share the implementation with the calling agent.
This effectively allows us to utilize more powerful models such as GPT-5-Pro to generate code for us or entire implementations (which are either API-only or part of the $200 Pro plan from within the ChatGPT app)
