A collection of 130+ specialized Codex subagents covering a wide range of development use cases.
See the code
This repository serves as the definitive collection of Codex Subagents, specialized AI assistants designed for specific development tasks. Written specifically for Codex and aligned with the official docs.
Use Codex custom agent directories exactly as documented:
~/.codex/agents/ for global agents (available in all projects).codex/agents/ for project-specific agents (higher precedence in that repo).toml agent files you want into one of the directories above.Examples:
mkdir -p ~/.codex/agents
cp categories/01-core-development/backend-developer.toml ~/.codex/agents/
mkdir -p .codex/agents
cp categories/04-quality-security/reviewer.toml .codex/agents/
If you use agent configuration in Codex, keep it in .codex/config.toml under [agents] as described in the official docs.
 | Crawlbase is web data infrastructure trusted by 70,000+ developers. Its Crawling API, MCP server, and integrations give AI agents live access to any webpage β with JavaScript rendering, proxy rotation, and anti-bot protection. |
 | SerpApi is a Web Search API for your AI apps. Available in Markdown and JSON for any integration. |
π You can feature your product here and reach developers using AI coding agents like Claude Code, Codex, Gemini, and more. |
| Type | Path | Availability | Precedence |
|---|---|---|---|
| Project Subagents | .codex/agents/ | Current project only | Higher |
| Global Subagents | ~/.codex/agents/ | All projects | Lower |
Note: When naming conflicts occur, project-specific subagents override global ones.
Each subagent uses a Codex-native .toml format:
name = "subagent-name"
description = "When this agent should be invoked"
model = "gpt-5.6-luna"
model_reasoning_effort = "medium"
sandbox_mode = "read-only"
[instructions]
text = """
You are a [role description and expertise areas]...
[Agent-specific checklists, patterns, and guidelines]...
"""
Each subagent includes a model field that automatically routes it to the right model -- balancing quality and cost:
| Model | When It's Used | Examples |
|---|---|---|
gpt-5.6-sol | Deep reasoning -- architecture reviews, security audits, financial logic | security-auditor, architect-reviewer, fintech-engineer |
gpt-5.6-terra | Implementation, diagnostics, evaluation, and multi-step professional analysis | docker-expert, test-automator, scientific-literature-researcher |
gpt-5.6-luna | Bounded search, extraction, routing, drafting, and lightweight synthesis | search-specialist, docs-researcher, agent-installer |
Each subagent's sandbox_mode field controls filesystem access:
sandbox_mode = "read-only" - analyze without modifyingsandbox_mode = "workspace-write" - create and modify filesEssential development subagents for everyday coding tasks.
Language-specific experts with deep framework knowledge.
DevOps, cloud, and deployment specialists.
Subagents are specialized AI assistants that enhance Codex's capabilities by providing task-specific expertise. They act as dedicated helpers that Codex can call upon when encountering particular types of work.
Independent Context Windows Every subagent operates within its own isolated context space, preventing cross-contamination between different tasks and maintaining clarity in the primary conversation thread.
Domain-Specific Intelligence Subagents come equipped with carefully crafted instructions tailored to their area of expertise, resulting in superior performance on specialized tasks.
Shared Across Projects After creating a subagent, you can utilize it throughout various projects and distribute it among team members to ensure consistent development practices.
Explicit Delegation Codex does not spawn subagents automatically. Use explicit delegation prompts to specify which agents to spawn, how to divide the work, and what shape the result should take.
.toml agent files aligned with official Codex subagent docsPR review workflow:
Review this branch with parallel subagents. Have reviewer look for correctness, security, and missing tests. Have docs_researcher verify the framework APIs this patch depends on. Wait for both and summarize the findings with file references.
Bug investigation workflow:
Investigate the broken settings flow. Have code_mapper trace the owning code paths, browser_debugger reproduce the bug in the browser, and frontend_developer propose the smallest fix after the failure is understood. Wait for the read-heavy agents first, then continue.
Repo exploration and planning workflow:
Use search_specialist to locate the code related to payment retries, knowledge_synthesizer to summarize the current design, and refactoring_specialist to propose a minimal refactor plan. Return a concrete action list.
You ship products with AI, but every launch still dies quietly because nobody posts about it. EveryFeed plugs your AI assistant into a social workspace that drafts, schedules, and publishes across 35+ channels β no agency, no marketing hire.
We welcome contributions! See CONTRIBUTING.md for guidelines.
MIT License - see LICENSE
This repository is a curated collection of subagent definitions contributed by both the maintainers and the community. All subagents are provided "as is" without warranty. We do not audit or guarantee the security or correctness of any subagent. Review before use, the maintainers accept no liability for any issues arising from their use.
If you find an issue with a listed subagent or want your contribution removed, please open an issue in this repository and we'll address it promptly.
(top 24 of 29)
1,992 followers Β· starred Mar 2026
2,549 followers Β· starred Apr 2026
475 followers Β· starred Mar 2026
609 followers Β· starred Mar 2026
A collection of 130+ specialized Codex subagents covering a wide range of development use cases.
See the code
This repository serves as the definitive collection of Codex Subagents, specialized AI assistants designed for specific development tasks. Written specifically for Codex and aligned with the official docs.
Use Codex custom agent directories exactly as documented:
~/.codex/agents/ for global agents (available in all projects).codex/agents/ for project-specific agents (higher precedence in that repo).toml agent files you want into one of the directories above.Examples:
mkdir -p ~/.codex/agents
cp categories/01-core-development/backend-developer.toml ~/.codex/agents/
mkdir -p .codex/agents
cp categories/04-quality-security/reviewer.toml .codex/agents/
If you use agent configuration in Codex, keep it in .codex/config.toml under [agents] as described in the official docs.
| Crawlbase is web data infrastructure trusted by 70,000+ developers. Its Crawling API, MCP server, and integrations give AI agents live access to any webpage β with JavaScript rendering, proxy rotation, and anti-bot protection. |
| SerpApi is a Web Search API for your AI apps. Available in Markdown and JSON for any integration. |
π You can feature your product here and reach developers using AI coding agents like Claude Code, Codex, Gemini, and more. |
| Type | Path | Availability | Precedence |
|---|---|---|---|
| Project Subagents | .codex/agents/ | Current project only | Higher |
| Global Subagents | ~/.codex/agents/ | All projects | Lower |
Note: When naming conflicts occur, project-specific subagents override global ones.
Each subagent uses a Codex-native .toml format:
name = "subagent-name"
description = "When this agent should be invoked"
model = "gpt-5.6-luna"
model_reasoning_effort = "medium"
sandbox_mode = "read-only"
[instructions]
text = """
You are a [role description and expertise areas]...
[Agent-specific checklists, patterns, and guidelines]...
"""
Each subagent includes a model field that automatically routes it to the right model -- balancing quality and cost:
| Model | When It's Used | Examples |
|---|---|---|
gpt-5.6-sol | Deep reasoning -- architecture reviews, security audits, financial logic | security-auditor, architect-reviewer, fintech-engineer |
gpt-5.6-terra | Implementation, diagnostics, evaluation, and multi-step professional analysis | docker-expert, test-automator, scientific-literature-researcher |
gpt-5.6-luna | Bounded search, extraction, routing, drafting, and lightweight synthesis | search-specialist, docs-researcher, agent-installer |
Each subagent's sandbox_mode field controls filesystem access:
sandbox_mode = "read-only" - analyze without modifyingsandbox_mode = "workspace-write" - create and modify filesEssential development subagents for everyday coding tasks.
Language-specific experts with deep framework knowledge.
DevOps, cloud, and deployment specialists.
Subagents are specialized AI assistants that enhance Codex's capabilities by providing task-specific expertise. They act as dedicated helpers that Codex can call upon when encountering particular types of work.
Independent Context Windows Every subagent operates within its own isolated context space, preventing cross-contamination between different tasks and maintaining clarity in the primary conversation thread.
Domain-Specific Intelligence Subagents come equipped with carefully crafted instructions tailored to their area of expertise, resulting in superior performance on specialized tasks.
Shared Across Projects After creating a subagent, you can utilize it throughout various projects and distribute it among team members to ensure consistent development practices.
Explicit Delegation Codex does not spawn subagents automatically. Use explicit delegation prompts to specify which agents to spawn, how to divide the work, and what shape the result should take.
.toml agent files aligned with official Codex subagent docsPR review workflow:
Review this branch with parallel subagents. Have reviewer look for correctness, security, and missing tests. Have docs_researcher verify the framework APIs this patch depends on. Wait for both and summarize the findings with file references.
Bug investigation workflow:
Investigate the broken settings flow. Have code_mapper trace the owning code paths, browser_debugger reproduce the bug in the browser, and frontend_developer propose the smallest fix after the failure is understood. Wait for the read-heavy agents first, then continue.
Repo exploration and planning workflow:
Use search_specialist to locate the code related to payment retries, knowledge_synthesizer to summarize the current design, and refactoring_specialist to propose a minimal refactor plan. Return a concrete action list.
You ship products with AI, but every launch still dies quietly because nobody posts about it. EveryFeed plugs your AI assistant into a social workspace that drafts, schedules, and publishes across 35+ channels β no agency, no marketing hire.
We welcome contributions! See CONTRIBUTING.md for guidelines.
MIT License - see LICENSE
This repository is a curated collection of subagent definitions contributed by both the maintainers and the community. All subagents are provided "as is" without warranty. We do not audit or guarantee the security or correctness of any subagent. Review before use, the maintainers accept no liability for any issues arising from their use.
If you find an issue with a listed subagent or want your contribution removed, please open an issue in this repository and we'll address it promptly.
(top 24 of 29)
1,992 followers Β· starred Mar 2026
2,549 followers Β· starred Apr 2026
475 followers Β· starred Mar 2026
609 followers Β· starred Mar 2026