Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
There was a problem hiding this comment.
Pull request overview
Adds a new Claude CLI-backed agent and begins shifting agent I/O toward writing structured JSON to a response file (instead of stdout), with shared JSON parsing utilities and related prompt/executor updates.
Changes:
- Introduces
ClaudeCLIAgentand registers it in the agent factory. - Adds shared JSON response parsing utilities and refactors Gemini CLI parsing to use them.
- Updates system prompts and execution flow in preparation for response-file based outputs.
Reviewed changes
Copilot reviewed 8 out of 8 changed files in this pull request and generated 10 comments.
Show a summary per file
| File | Description |
|---|---|
| src/mergai/prompts/system_prompt_resolve.md | Changes output instructions to response-file/Write-tool based JSON output. |
| src/mergai/prompts/system_prompt_describe.md | Changes output instructions to response-file/Write-tool based JSON output. |
| src/mergai/app.py | Forces YOLO for describe() to support Claude CLI writing a response file. |
| src/mergai/agents/response_utils.py | Adds shared helpers to extract/parse JSON from agent outputs. |
| src/mergai/agents/gemini_cli.py | Refactors response parsing toward shared utilities and response-file support (currently broken). |
| src/mergai/agents/factory.py | Registers the new claude-cli agent type. |
| src/mergai/agents/claude_cli.py | Adds the Claude CLI agent implementation (stream-json parsing, session reading, optional response-file read). |
| src/mergai/agent_executor.py | Moves prompt files into state dir and starts adding response-file path handling/validation flow (currently broken). |
Comments suppressed due to low confidence (2)
src/mergai/agents/gemini_cli.py:189
- This block still calls
fix_response_json(...), but that helper was removed from this file and is no longer imported. It will crash at runtime; it looks like the intent is to useparse_response_json(result)(or skip parsing if the response was read from a file).
try:
result = fix_response_json(response)
version = self.get_version()
except AgentError as e:
src/mergai/agent_executor.py:191
- A
response_pathis created, but it’s never passed to the agent and the prompt never mentions it, so the updated system prompts (“file path will be provided”) can’t be satisfied. If response-file mode is required, include the path incurrent_promptand pass it to agents that support it (or update the baseAgent.runinterface to acceptresponse_file).
# Create response file path in state_dir (same location as prompt)
# state_dir is typically .cache/mergai which is gitignored
response_filename = self._generate_response_filename()
response_path = self.state_dir / response_filename
current_prompt = (
f"See @{prompt_path} make sure the output is in specified format"
)
error = None
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated 8 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated 2 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
plebioda
added a commit
that referenced
this pull request
Apr 1, 2026
- gemini_cli.py: Add fallback to stdout parsing when response file is missing or contains invalid JSON, instead of failing immediately - agent_executor.py: Fix retry loop to execute all attempts (was breaking before final attempt). Preserve file instructions across retries by appending error context to base prompt instead of replacing it. - response_utils.py: Add explicit type check for response before calling extract_json_block() to avoid TypeError from re.search on non-strings. Also simplified exception handling to only catch JSONDecodeError.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
- Add ClaudeCLIAgent for integration with Claude Code CLI - Support JSON streaming output and session management - Support both yolo mode and limited write permissions (acceptEdits) - Parse token usage statistics and cost information - Implement file-based response pattern across all agents - Agents write JSON responses to designated files in state_dir - Add response_file and allowed_write_paths parameters to Agent.run() - Provides more reliable JSON parsing than stdout extraction - Extract shared response parsing utilities to response_utils.py - Move JSON block extraction and parsing from gemini_cli.py - Support both direct JSON and markdown-wrapped formats - Update system prompts to instruct agents to write to response files - Improve validation error messages to list modified files
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.