policy-starter

Draft a firm AI usage policy from published model policies, adapted to your practice profile — a research-and-synthesis tool whose output is a draft for attorney review and adoption, not a finished policy. Use when user says "draft an AI policy", "we need an AI policy", "build an AI usage policy", "our firm needs a GenAI policy", or similar requests to generate a first-cut internal AI policy.

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

85%Weight 40%Scale 1-3

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a well-structured, highly actionable instruction set with a clear sequenced workflow and explicit validation checkpoints. Its only weakness is mild verbosity from overlapping rationale across the Purpose, top-level steps, and Don'ts sections.

Suggestions

Collapse the 'Purpose' discipline list into the top 7-step procedure or the Don'ts to remove the overlap and tighten the body toward a leaner token budget.

The 'Output scope reminder' near the end reiterates the header-audience point already covered in the Header subsection; consider merging it into that subsection.

Trim restated rationale (e.g., 'A policy that tries to cover everything covers nothing' appears in both the Purpose and Don'ts sections) to a single canonical location.

Dimension	Reasoning	Score
Conciseness	The ~250-line body is mostly efficient domain-specific guidance rather than generic concept padding, but the 'Purpose' discipline list overlaps the top 7-step procedure and the 'Don'ts' section, so it could be tightened — matching anchor 2 rather than the lean anchor 3.	2 / 3
Actionability	Provides copy-paste-ready instructional artifacts — a 10-item scope-interview checklist with a default starter pack, a jurisdiction-to-source mapping table, an exact header template, a sources-block and reviewer-note block format, and an adoption checklist — matching anchor 3 for an instruction-only skill.	3 / 3
Workflow Clarity	A clear sequenced workflow (read profile → scope interview → source model policies → draft → output → handoffs) with explicit validation checkpoints — stop if profile unpopulated, push back if user skips scope, ask if footprint is a placeholder, note if web search fails — matching anchor 3.	3 / 3
Progressive Disclosure	No bundle files exist, so all content lives in SKILL.md, but it is organized into well-signaled sections (Matter context, Purpose, Scope interview, The draft, Don'ts, Handoffs) with no nested references, which satisfies the simple-skill allowance for anchor 3.	3 / 3
	Total	11 / 12 Passed

Description

100%Weight 40%Scale 1-3

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is strong: it states a concrete capability, includes an explicit 'Use when' clause with several natural trigger phrasings, and carves out a distinct legal-policy-drafting niche. It uses third-person/neutral voice with no over-claims, so it meets the top anchor on every dimension.

Dimension	Reasoning	Score
Specificity	Names multiple concrete actions/characteristics — 'Draft a firm AI usage policy from published model policies, adapted to your practice profile' and frames it as 'a research-and-synthesis tool whose output is a draft for attorney review'. It is more conceptual than the extract/fill/merge example but still lists distinct concrete actions, so it clears anchor 3 rather than sitting at 2.	3 / 3
Completeness	Explicitly answers both what ('Draft a firm AI usage policy from published model policies... output is a draft for attorney review and adoption, not a finished policy') and when ('Use when user says...'), with an explicit 'Use when' trigger clause, matching anchor 3 and not the what-only anchor 2.	3 / 3
Trigger Term Quality	Provides natural user phrasings — 'draft an AI policy', 'we need an AI policy', 'build an AI usage policy', 'our firm needs a GenAI policy' — covering common variations a user would actually say, matching anchor 3 rather than the partial coverage of anchor 2.	3 / 3
Distinctiveness Conflict Risk	A clear niche (drafting an internal AI usage policy for a legal/firm context) with distinct GenAI-policy triggers makes it unlikely to fire for unrelated skills, matching anchor 3.	3 / 3
	Total	12 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: anthropics/claude-for-legal
Path: ai-governance-legal/skills/policy-starter/SKILL.md
Commit: 4a6c651

Reviewed: 6 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.