Is it safe to use on sensitive codebases?

Yes, it features strict safety rules: it redacts secrets from configurations, prevents subagent recursion, and offers a 'read-only' sandbox mode for analysis-only tasks.

How do I prevent timeouts during long-running subagent tasks?

You should increase the tool_timeout_sec in your config.toml to 600 seconds for the codex-specialized-subagents server to accommodate complex delegated runs.

Can I control the number of subagents used?

Yes, you can specify the 'max_agents' parameter (default is 3) to limit how many subagents are deployed for a single task.

When does Delegation Autopilot trigger automatically?

It triggers when a request is multi-step (using keywords like 'and' or 'then'), touches multiple areas like code and tests, or involves cross-cutting changes across several files.

Delegation Autopilot

Name: Delegation Autopilot
Author: hou27

byhou27

Productivity & Workflow

Automates the orchestration of specialized subagents for complex, multi-step coding tasks and cross-cutting repository changes.

About

Delegation Autopilot transforms the Claude parent agent into a project manager capable of identifying complex requests that require multi-stage execution or specialist attention. By automatically triggering subagents for tasks involving concurrent code, test, and documentation updates, it ensures consistent 'scan-implement-verify' workflows while maintaining a safe, controlled sandbox environment for large-scale repository modifications and deep research.

Key Features

Consolidated aggregation of results and artifacts from all subagent runs
0 GitHub stars
Configurable sandbox modes including workspace-write and read-only analysis
Intelligent orchestration of specialized subagents for cross-cutting changes
Automatic detection and decomposition of multi-step user requests
Built-in recursion protection and sensitive data redaction for security

Use Cases

Performing repository-wide research and analysis with structured reporting
Implementing new features that require simultaneous updates to logic, unit tests, and documentation
Executing complex refactors that touch multiple modules or files across a codebase

About

Key Features

Consolidated aggregation of results and artifacts from all subagent runs
0 GitHub stars
Configurable sandbox modes including workspace-write and read-only analysis
Intelligent orchestration of specialized subagents for cross-cutting changes
Automatic detection and decomposition of multi-step user requests
Built-in recursion protection and sensitive data redaction for security

Use Cases

Performing repository-wide research and analysis with structured reporting
Implementing new features that require simultaneous updates to logic, unit tests, and documentation
Executing complex refactors that touch multiple modules or files across a codebase