define-agent

Part of the ACED plugin — see that page for install instructions.

Trigger: “create an agent”, “write a conductor”, “make an agent definition”, “define a persona”, “build a dual-mode agent”, “add an invokable agent”

Creates or improves an agent definition — a named, reusable role encoded in a single file.

Agent definition modes

Mode	What it does	When to pick it
Delegated	Runs as a subagent in its own context; returns a result to the caller	Autonomous workers, fan-out tasks
Invokable (dual-mode)	Spawnable as a subagent AND loadable in-context via a companion command	Conductors, reviewers, personas the user steers interactively
In-context only	Loaded via command only, not a subagent	Short personas, voice/register adopters

Invokable mode scaffolds a companion command file alongside the agent definition, sharing one body.

What it does

1. Determines placement (user-global, project, or inside a plugin) and target runtimes (Claude Code, Cursor, Codex), then derives the canonical path and symlinks.
2. Gathers name, role, responsibilities, output format, human-in-the-loop rules, out-of-scope, and tools.
3. Drafts the agent definition, and for Invokable mode, the companion command.
4. Runs quality checks (frontmatter shape, description trigger phrasing, body structure) and fixes CRITICAL/HIGH failures before presenting the file.

Role as ACED impl-producer

When the SDD conductor dispatches this skill against a frozen .feature, it co-produces the agent definition and its verification — the eval suite (the frozen .feature’s scenarios plus the node’s eval.md run policy) that the impl-judge runs. Invoked standalone (no frozen .feature), it produces only the agent definition.

Next step

Run sdd:start-mission to spec and eval the agent definition; the conductor resolves the ACED roles automatically.