Skill Authoring

Skill Authoring

Repository continuity composition

For repository-scoped work, compose maintain-repository-continuity in Resume mode before selecting work. After an authorized repository change passes domain validation, compose Refresh and Verify immediately before presenting the pull request, and include the reconciled root CONTINUITY.md in the same change. A policy-permitted no-change or exemption result must be documented instead of fabricating an edit.

• Contribute: Skill identity, package changes, version, provenance, evaluations, validation, and compatibility
• Never claim: That a skill was installed, released, adopted, or made authoritative without evidence

Purpose

Create or maintain a portable skill package whose instructions, resources, metadata, provenance, and evaluations are reusable across repositories and hosts.

Required Inputs

Resolve as much of the following as possible:

• requested capability and trigger conditions
• canonical output path and stable skill identity
• repository or organization conventions that the skill must respect
• required references, templates, or scripts
• provenance, authorship, and license status
• validation and evaluation expectations
• open questions or unresolved ownership boundaries

Record missing evidence instead of inventing conventions.

Workflow

1. define the reusable capability and its routing boundary

2. choose one stable kebab-case identity and canonical directory

3. normalize frontmatter to the Agent Skills and Aether contracts

4. keep the core workflow concise and move detail into focused resources such as:

   • ./references/validation-checklist.md
   • ./templates/SKILL.template.md

5. add only the resources that reduce repeated work or make execution deterministic

6. make authorship, provenance, license, and adoption status explicit

7. create deterministic eval coverage for trigger, boundary, insufficient-evidence, negative, and update scenarios

8. validate links, package structure, metadata, and any included executable resources

Constraints

• Do not turn a skill into a generic agent or plugin installer.
• Do not duplicate repository-wide instructions that belong elsewhere.
• Do not keep synonym identities alive without a clear behavior split.
• Do not add scripts, assets, or examples that do not reduce repeated work.
• Do not claim first-party ownership without explicit evidence.

Completion Criteria

• Stable identity and canonical path are explicit.
• Frontmatter satisfies the current skill contract.
• Progressive-disclosure resources are focused and necessary.
• Provenance, license, and source-delta decisions are visible.
• Deterministic eval coverage exists for the supported workflow.
• Validation steps and remaining questions are reported truthfully.

Provenance

This canonical skill is first-party Ego Hygiene content curated from the staged candidate at .staging/skills/skill-authoring/SKILL.md.

Source Delta

• Adopted: the staged workflow for defining capability, planning resources, keeping SKILL.md concise, and validating package structure.
• Rewritten: canonical frontmatter, Aether metadata, reusable template, evaluation coverage, and explicit provenance handling.
• Rejected: the staged .agents/skills/ placement guidance from the historical copy in favor of the canonical library/organization/skills/<domain>/<name>/ source layout; the broader create-skill synonym remains only as a reviewed legacy reference.