What problem does it solve?
Writing agent skills that comply with the Anthropic and agentskills.io specification, activate correctly, and perform reliably in production is error-prone. Common pitfalls like invalid frontmatter, oversized body content, nested references, and missing evaluation steps lead to skills that fail to trigger or produce inconsistent results.
Core Features & Use Cases
- Spec-compliant authoring guidance: Covers frontmatter rules, body budget limits, and reference layout requirements to ensure skills pass validation and activate as expected.
- Evaluation-first workflow: Includes a structured loop to baseline agent performance without the skill, write test cases, and iterate based on real execution traces to close capability gaps.
- Pre-publish linting: Bundles a lint script to catch errors like reserved name substrings, first-person descriptions, and broken reference links before shipping.
- Use Case: For example, if you are building a skill to automate your team's Slack message triage workflow, this skill walks you through writing the SKILL.md, structuring reference documentation, bundling helper scripts, and validating the skill to ensure it activates when users mention "triage Slack messages" and performs the task correctly.
Quick Start
Use the authoring-skills skill to create a new skill that automates your team's weekly status report generation and posting to the internal wiki.
