Skip to content
Back to skills

Add Agent Definition

ASecurity

Procedural workflow for adding support for a new coding agent to agenteq.

  • 6 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 19, 2026
ai-agentstypescript

Works with

  • cli
  • mcp

Security analysis

A100/100

Scanned September 19, 2026

npx -y skills add sunchayn/agenteq --skill add-agent-definition --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Add Agent Definition?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Add Agent Definition
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/sunchayn-add-agent-definition/badge)](https://www.skillsdirectory.com/skills/sunchayn-add-agent-definition)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: add-agent-definition

description: Procedural workflow for adding support for a new coding agent to agenteq.
---

# Adding an Agent Definition

## 1. What an agent definition is

One file under `src/modules/agents/definitions/`, exporting a single `new Agent({...})`. A capability that needs nothing beyond a directory or file path, `guidelinesPath`, `commandsDir`, `skillsDir`, is a bare string field. A capability with more shape to configure, `mcp`, gets its own configuration object, built with a `create{Name}Configuration` factory. `definitions/claude-code.ts` is a filled-in reference for this shape.

## 2. Procedural steps

1. Create `src/modules/agents/definitions/{name}.ts`, default-exporting one `new Agent({...})`. See [write-typescript-code](../write-typescript-code/SKILL.md) section 4 for why this is a default export.
2. Set `name` (machine-readable, snake_case) and `displayName` (human-readable).
3. Set `detectSystemPathUsing`, a closure returning a `DetectionConfiguration` for how the agent's CLI is detected on the system, usually a `command` check.
4. Set `detectProjectPathUsing`, a closure returning a `DetectionConfiguration` for how the agent is detected as already configured in the current project, usually `files` or `paths` checks.
5. Opt into only the capabilities the agent actually supports: `guidelinesPath`, `commandsDir`, `skillsDir` (a plain string path each), `mcp` (a `McpConfiguration` built with `createMcpConfiguration`, taking a `configPath` and `entryMappers`).
6. Register it in `src/modules/agents/registry.ts`: one import, one entry added to the exported `agentDefinitions` array.

No other source file should need to change. If adding an agent requires touching a sync stage, that's a sign a capability abstraction is leaking agent-specific logic. Reconsider the capability configuration instead of special-casing the new agent inside a stage.

Two files outside the source tree still need updating: [refactor-to-agenteq](../refactor-to-agenteq/SKILL.md) and [add-agenteq-to-boost-project](../add-agenteq-to-boost-project/SKILL.md) each carry an identical reference table hardcoding every agent's native guidelines path, skills/commands directory, command file extension, and MCP config path, key path, and entry shape. Those skills run in a different repo with no access to this codebase, so neither can read `definitions/{name}.ts` directly. Adding a new agent, or changing any of those fields on an existing one, needs the matching row added or updated in both tables.

## 3. Related Skills

- **General TypeScript conventions**: [write-typescript-code](../write-typescript-code/SKILL.md)
- **Adding a sync capability**: [add-capability-stage](../add-capability-stage/SKILL.md)
- **Verifying the new definition's output**: [e2e-test-agents](../e2e-test-agents/SKILL.md)
- **Keeping the reverse-engineering tables current**: [refactor-to-agenteq](../refactor-to-agenteq/SKILL.md), [add-agenteq-to-boost-project](../add-agenteq-to-boost-project/SKILL.md)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.