skills.rest
RavitejaKarra24
RavitejaKarra24COMMUNITY
Explore all skills by this author →
𝕏@RavitejaKarra24GitHub

subagents

Spawn and manage headless subagents across pi, Claude Code, and Codex harnesses.

Updated Mar 12, 2026
Search Tags:#claude code#ai agents#parallel execution#subagents#task delegation#model selection#codex cli
One-click install
npx skills add https://github.com/RavitejaKarra24/dotfiles --skill subagents-ravitejakarra24
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagents
Source: https://github.com/RavitejaKarra24/dotfiles/tree/main/pi/.pi/agent/skills/subagents
Command: npx skills add https://github.com/RavitejaKarra24/dotfiles --skill subagents-ravitejakarra24

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Delegating work to subagents requires knowing which harness to use, which models and thinking budgets each one supports, and how to spawn, monitor, and collect results without blocking the parent conversation. ## Core Features & Use Cases - Multi-Harness Spawning: Launch headless subagents on the pi, Claude Code, or Codex harness, each with its own context window and a self-contained prompt. - Model and Effort Selection: Pick recommended models and thinking budgets per harness, such as gpt-5.6-sol on high effort for Codex or the latest Claude Fable for Claude Code. - Lifecycle Management: Check, list, wait on, or cancel up to four concurrent runs, with results returning automatically to the parent. - Use Case: Ask the agent to run three independent code reviews in parallel as Codex subagents while the parent continues editing, then collect their reports when ready. ## Quick Start Spawn a Codex subagent to review the authentication module and report any security issues.

Frequently Asked Questions about subagents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I spawn a subagent with a specific model?▼

Call subagent_spawn with a complete prompt, a short name, the chosen harness, and optional model and reasoning_effort parameters. Use provider/model-id format like openai-codex/gpt-5.6-sol for unambiguous model selection.

What is the difference between pi, Claude Code, and Codex subagent harnesses?▼

The pi harness inherits the parent model and thinking level by default. Claude Code defaults to the latest Claude Fable model on high effort, while Codex defaults to gpt-5.6-sol on high effort for coding work.

How many subagents can run at the same time?▼

At most four subagents run concurrently. You can track them with subagent_list, peek at progress with subagent_check, and block on results with subagent_wait only when needed to proceed.

Can a subagent ask the user questions or spawn its own subagents?▼

No. Each subagent is headless, cannot see the parent conversation, cannot ask the user, and cannot spawn subagents or workflows. Every child prompt must be self-contained with paths, constraints, and the expected report.

What thinking budgets do subagents support?▼

All harnesses accept off, minimal, low, medium, high, xhigh, and max. Claude Code maps these to token budgets from 0 to 63,999, while Codex maps them to the nearest effort supported by the selected model.

Related Skills

Explore More Software Engineering Skills→
greglambgreglamb
1

dispatching-parallel-agents

Decompose independent engineering tasks into parallel agent assignments with conflict review.

Community
Intermediate
Piece-PuzzlyPiece-Puzzly

review-fix

Apply CodeRabbit PR review comments to code and push changes.

Official
Advanced
Euda1mon1aEuda1mon1a

automated-code-fixer

Detect and correct code defects across test suites, linters, type checkers, and security scanners.

Community
Advanced
CasaPerksCasaPerks

sdd

Execute implementation plan tasks via isolated subagents with spec-first review.

Official
Advanced