skill-design-patterns
PassAudited by ClawScan on May 12, 2026.
Overview
This is an instruction-only skill-design guide with no code or credentials; its only notable behavior is recommending optional project-wide agent guidance that should be reviewed before making persistent.
This looks safe to install as guidance-only. If you copy its recommendations into AGENTS.md or another persistent configuration, review them first and keep them scoped to the intended project or team. The provided SKILL.md excerpt was truncated, so skimming the full file before installation is still prudent.
Publisher note
Guides the design of effective agent skills. Use when creating a new skill, improving an existing skill, reviewing a skill's structure, or establishing skill-writing conventions for a project. Covers anti-rationalization tables, red flags, verification checklists, core operating behaviors, and progressive disclosure.
Findings (1)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If these recommendations are copied into AGENTS.md or a similar persistent file, they may influence future agent behavior across the project or team.
This is guidance to place persistent instructions into a project or platform entry-point file. It is disclosed and aligned with establishing skill-writing conventions, but persistent instructions can affect future agent behavior beyond the immediate task.
These six non-negotiable behaviors should be defined in your platform's entry-point configuration (AGENTS.md or equivalent). They apply across all skills and sessions.
Review any proposed AGENTS.md or persistent configuration changes before applying them, and scope them to the intended project or team.