| name | agent-spec-creation |
| description | Create or update agent integration specs before implementation. Use this whenever a request is about defining how a new or existing agent should behave in Ralph. |
Agent Spec Creation Workflow
Purpose
Create implementation-ready agent integration specifications without changing production code.
When to Use
Use this skill when the task is to define or revise agent behavior, inputs, outputs, error handling, or acceptance criteria in specs.
Required Workflow
- Read intent and existing spec context first:
specs/README.md
specs/agents.md
- relevant files in
specs/agents/*.md
- Rely on the
spec-creator skill for document structure and quality requirements.
- Study the target CLI behavior before writing spec details:
<agent-cli> --help
- relevant subcommand help (for example
<agent-cli> run --help)
- Define explicit, testable requirements:
- command shape and required flags
- environment variables
- output parsing/streaming expectations
- non-interactive execution behavior
- failure modes and expected surfaced errors
- Update spec files only.
Constraints
- Do not implement code in
internal/agent/* while using this skill.
- Do not run iterative Ralph loop automation.
- Keep the workflow linear: read, specify, and stop after spec updates.
Completion Criteria
- Spec changes are clear, deterministic, and testable.
- Scope and non-goals are explicit.
- Related index files (for example
specs/README.md) are updated when new spec files are added.