agent-dispatcher
Repository retrieval and context engine for Claude Code and Codex: finds relevant code, reuses project knowledge, and pairs each task with specialist skills and verification.
50stars
4forks
Python
Added 9/24/2026
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
AI Agentsagent-skillsai-agentsautomationclaude-codecode-reviewdeveloper-toolsmcpmulti-agentprompt-engineeringtesting
Installation
# Add to your Claude Code skills
git clone https://github.com/nahid-sparktales/agent-dispatcherGetting Started
Guides for using ai agents skills like agent-dispatcher.
- Caveman: Cut Claude Token Use by 65%How agent-side prompt compression works, when to use it, and when not to.
- What is an AI Skills Marketplace?Definitions, how marketplaces work, and how to choose between them in 2026.
- Getting Started with AI SkillsFirst-time install walkthrough for Claude Code, Codex CLI, and ChatGPT.
Compare with similar skills
See how agent-dispatcher compares with popular alternatives.
Agent Dispatcher
Repository retrieval and context for Claude Code and Codex.
Agent Dispatcher finds the code a task needs using symbols, source text, code relationships, and Git history. It assembles relevant source excerpts, pairs the work with specialist guidance, and makes verification explicit. Use it to explore a repository, debug a failure, build a feature, or review a change.
Quick start · Usage · Catalog · Documentation · Changelog
Quick start
You need an installed, authenticated Claude Code or Codex, Git, and Python 3.10+. The helpers use Python's standard library. Use macOS, Linux, or a Unix environment such as WSL; native Windows installation is not covered. Normal use needs no additional API key.
Codex
Run in your terminal:
git clone https://github.com/nahid-sparktales/agent-dispatcher.git
cd agent-dispatcher
python3 install_codex.py
Start a new Codex task in your project and invoke the skill:
$agent-dispatcher Find why the tests are failing, fix the cause, and verify the fix.
This installs one skill at ~/.agents/skills/agent-dispatcher/, with roles and supporting
guides loaded as needed. See Codex setup for updates and uninstall.
Claude Code
Run in your terminal:
claude plugin marketplace add nahid-sparktales/agent-dispatcher
claude plugin install agent-dispatcher@agent-dispatcher
Start a new Claude Code session in your project:
/agent-dispatcher:agent-dispatcher Find why the tests are failing, fix the cause, and verify the fix.
Prefer a manual install? Follow the manual setup guide. Use one installation method per host to avoid duplicate commands and hooks.
Everyday use
In Codex, add your request or a control after $agent-dispatcher:
$agent-dispatcher Redesign the settings page and check it on mobile.
$agent-dispatcher reviewer Review this change for correctness and missing tests.
$agent-dispatcher debugger Find why checkout fails after login.
Claude plugin users use /agent-dispatcher:agent-dispatcher for routing and controls, or a
role command such as /agent-dispatcher:agent-reviewer. These prefixes follow
Claude's plugin command names.
Manual installs use /agent-dispatcher and /agent-reviewer.
| Add after the dispatcher invocation | What it does |
|---|---|
context explain |
Explain the current role, skills, and context selection. |
context build <request> |
Find relevant source passages without executing the request. |
doctor |
Check installation health and recommend relevant setup. |
inventory setup |
List capabilities with missing setup or unresolved availability. |
output verbose / output compact |
Change how much activity detail is shown. |
status |
Show activation state and current settings. |
on here / off here |
Enable or silence automatic activation for this project. |
Automatic activation is opt-in. Codex requires you to review and trust its hook through
/hooks. A chosen role stays active until you select another or stop routing.
See all commands and configuration.
What it does
- Finds relevant code. Retrieval combines paths, symbols, source text, stack-trace frames, code relationships, and Git history. A bounded context packet includes actual source excerpts and the reasons each file was selected.
- Reuses project knowledge. Source-linked project maps and private caches reuse unchanged evidence. An explicitly built deep index adds broader repository coverage, resumable onboarding, and incremental refresh.
- Can use task experience. Explicitly recorded outcomes, verification receipts, and corrections share one private store. Repository memory can also use eligible commit history and module summaries, with freshness checks before they influence retrieval.
- Matches the work to a specialist. Roles cover engineering, design, testing, research, writing, and operations. Small, obvious edits take a direct path; substantial tasks get specialist guidance. Independent subtasks can use the host's subagents when authorized.
- Makes checks visible. Verification receipts distinguish passed checks, failures, zero tests, and work left unverified. Optional change audits report which files actually changed. Replies default to short, plain language; saved preferences control presentation.
- Adapts to available tools. Inventory and health checks distinguish usable capabilities, missing setup, and unknown connections. Referenced external skills and MCP integrations have documented fallbacks.
Roles are instructions your coding agent adopts. Claude Code or Codex supplies the model, tools, and permission controls; Dispatcher respects the host's model and effort settings. Installing the pack does not connect external services or start background agents.
Defaults and optional features
Standard retrieval runs locally without model or network calls. Project indexes and memory stay in private storage outside the working tree. Model-assisted features require your own configuration and may send task or source information to the provider you select.
| Feature | Default and next step |
|---|---|
| Deep repository index | Build explicitly; later context preparation uses the published index automatically. |
| Task experience | Recording and retrieval enabled; records must be supplied explicitly. No records means no effect. |
| History and semantic memory | Build explicitly; retrieval starts in shadow mode, reporting candidates without changing rankings. |
| Procedural learning | Off; optionally review recorded outcomes into evaluated, human-approved overlays on bundled guidance, starting in shadow mode. |
| Model-assisted retrieval | Off; optionally generate file summaries and rerank a bounded candidate set. |
| Onboarding Explorer | Off; optionally use a model to investigate the index and save evidence-backed notes. |
| Jev decision engine | All scopes off; optionally delegate catalog selection to an external API. |
The catalog lists bundled skills and referenced integrations. External skills and MCP servers are not installed by this pack. Your agent's usage costs still apply; optional providers have their own access and billing requirements.
Documentation
| Start here | What you will find |
|---|---|
| Usage and configuration | Commands, output preferences, context controls, activation, and uninstall. |
| Catalog | Every role, local skill, workflow recipe, and MCP registry entry. |
| Repository intelligence | How files are found, ranked, and selected for context. |
| Project intelligence | Project maps, caching, and context budgets. |
| Benchmarks · End-to-end evaluations | Measured results, methodology, and limits. |
| Architecture · Verification · Security | Design, evidence requirements, and trust boundaries. |
| Changelog | Detailed changes and release history. |
Contributing
Bug reports, routing examples, and focused contributions are welcome. Read CONTRIBUTING.md for canonical sources and generated files, then run:
python3 build.py
python3 -B -m tests
These checks run offline without provider credentials. To extend the pack, see adding a role or adding a skill. Report vulnerabilities through SECURITY.md.
License
MIT. See NOTICE for role origins, adaptations, and attribution. Referenced third-party skills retain their own licenses. This is an independent project, not affiliated with or endorsed by OpenAI, Anthropic, or Locus.
Frequently Asked Questions
What is agent-dispatcher?
agent-dispatcher is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by nahid-sparktales. Repository retrieval and context engine for Claude Code and Codex: finds relevant code, reuses project knowledge, and pairs each task with specialist skills and verification. It has 50 GitHub stars.
Is agent-dispatcher safe to use?
agent-dispatcher's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
How do I install agent-dispatcher?
Clone the repository with "git clone https://github.com/nahid-sparktales/agent-dispatcher" and add it to your Claude Code skills directory (see the Installation section above).
What programming language is agent-dispatcher written in?
agent-dispatcher is primarily written in Python. It is open-source under nahid-sparktales on GitHub, so you can review or fork the full source.
Are there alternatives to agent-dispatcher?
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh agent-dispatcher against similar tools.
Comments (0)
to leave a comment.
No comments yet. Be the first to share your thoughts!
Related Skills
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
265,682
39,705
JavaScript
AI Agentsai-agentsanthropic
Skills for Real Engineers. Straight from my .agents directory.
250,518
21,171
Shell
AI Agents
An agentic skills framework & software development methodology that works.
234,966
20,863
Shell
AI Agentsaibrainstorming
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
205,748
60,855
TypeScript
MCP Serversaiapis