feat(dashboard): recent files list in Files view

[elecvoid243]

Implements docs/superpowers/specs/2026-07-20-recent-files-design.md. New (default-collapsed) above FileTreeList. Per-worktree 50-entry localStorage bucket; recordOpen on fileBrowserPreviewPath changes; row click mirrors onFileOpen; × button per row; Clear link with secondary-confirm dialog. No new deps. 28 new unit tests.

[sourcery-ai]

Sorry, we are unable to review this pull request

The GitHub API does not allow us to fetch diffs exceeding 300 files, and this pull request has 313

[gemini-code-assist]

Code Review

This pull request introduces comprehensive support for subagent orchestration, including a SubAgent Manager, a DAG orchestration engine, and associated tools (e.g., dynamic creation, task waiting, and shared context). It also enhances the local booter with Windows console suppression and host process termination protection, integrates a robust file editing engine with history rollback, and updates the dashboard with project indicators, a plan/build toggle, a draggable todo summary bar, and a CodeMirror editor. Feedback on the changes highlights a duplicate code block in the conversation reset command, a magic number that should be defined as a constant, duplicated subagent status-checking logic that should be refactored, and a regex-based process termination check that should be made more robust using shlex.split().

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

This block of code to clean up subagents is a duplicate of the one above (lines 193-203). This causes SubAgentManager.cleanup_session(umo) to be called twice and appends a second cleanup message to the result string, one in Chinese and one in English.

This is likely a copy-paste error and should be removed to avoid redundant operations and confusing output.

[gemini-code-assist]

The value 128000 is a magic number used as a fallback for max_context_tokens. To improve maintainability and readability, it's best to define this as a named constant at the top of the file or in a relevant constants module.

For example:

# At the top of the file
DEFAULT_MAX_CONTEXT_TOKENS = 128_000

# In this function
...
max_tokens = DEFAULT_MAX_CONTEXT_TOKENS

[gemini-code-assist]

The logic to check if a subagent is currently running is duplicated for transfer_to_subagent (lines 220-239) and for direct HandoffTool invocations (lines 256-275). This redundancy makes the code harder to maintain.

To avoid duplication, you could extract this check into a helper method within the FunctionToolExecutor class. For example:

    @classmethod
    def _check_and_yield_if_subagent_running(cls, run_context, agent_name: str | None):
        if not agent_name:
            return None
        if SubAgentManager.get_subagent_status(
            run_context.context.event.unified_msg_origin, agent_name
        ) == SubAgentStatus.RUNNING:
            return mcp.types.CallToolResult(
                content=[
                    mcp.types.TextContent(
                        type="text",
                        text=(
                            f"Error: SubAgent '{agent_name}' is currently RUNNING. "
                            "Wait for it to finish (e.g. via wait_for_subagent) or pick another subagent."
                        ),
                    )
                ]
            )
        return None

You could then call this helper in both places to simplify the execute method.

References

1. When implementing similar functionality for different cases (e.g., direct vs. quoted attachments), refactor the logic into a shared helper function to avoid code duplication.

[gemini-code-assist]

The regex used to detect kill commands with numeric PIDs is not robust enough. It can be bypassed with different argument formats, for example kill -s TERM 1234.

For better security, consider using shlex.split() to parse the command and then check for numeric arguments that match protected PIDs. This is more reliable than a regex for shell command parsing.

Example:

import shlex
...
    # In _would_kill_self function
    try:
        parts = shlex.split(lowered)
        if parts and parts[0] == 'kill':
            # This handles PIDs and negative PIDs (process groups)
            pids_in_cmd = [int(p) for p in parts if p.lstrip('-').isdigit()]
            if any(abs(pid) in protected for pid in pids_in_cmd):
                return True
    except ValueError:
        # Ignore commands with syntax errors like unclosed quotes
        pass

This would replace the current regex-based loop and provide stronger protection.

[elecvoid243]

神必loop事故请无视本PR