audience-generate-search

Build a person audience from a plain-English description — discover the signals behind the brief, score them, and compose, partition, or cross them into a measured result. The discovery-driven way into a build, behind /watt:audience; anchored on a description (no owned list). The objective picks how it lands — a size band, the widest credible reach, the highest-precision few, grouping into the best-concentrated cells, or crossing the employment graph to their employers (or employees). Produces a signal stack + measured reach, or a roster — never an export, never contact data. Not a user command. Use when a build starts from a description — "build me an audience of about 1–5M weekend hikers", "the highest-intent people in-market for solar", "where do these people cluster", "a lead list of in-market companies".

Quality

89%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

73%Weight 40%Scale 1-5

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

A densely specified, well-sequenced routing skill with strong workflow clarity and concrete, actionable guidance that correctly externalizes procedure bodies. Its main weakness is conciseness — the same guardrails are restated many times — and the body is heavier than a pure overview pointing at detailed materials.

Suggestions

Consolidate the repeated guardrails: state "membership only — ranking/segmenting the qualified set is a separate downstream strategy" and "read supported crossings from `entity_relations`, never an allow-list" once in How to behave and reference them, instead of restating each 3–5 times across the flow.

Move the per-objective breadth-lean detail (band/max-reach/precision/grouping/crossing) into a short table or a context file, keeping step 2 to the lean the leaf actually supplies and a one-line rationale per mode.

Verify or list the `context/*.md` files under a real bundle directory so the progressive-disclosure references can be confirmed to exist, and trim the SKILL.md toward an overview that defers more of the procedural narration to those files.

Dimension	Reasoning	Score
Conciseness	It assumes Claude's intelligence (no basic-concept padding), but repeats the same emphatic clarifications many times — "membership only — ranking or segmenting is a separate strategy" appears ~5×, "the supported crossings are whatever `entity_relations` exposes" 3×, and the "the target is the user's" theme is restated across both the flow and How to behave.	3 / 5
Actionability	Concrete and specific — exact `cat "<Bundle root>/context/<file>"` reads, validation bounds (`1,000 ≤ low < high ≤ 200,000,000`), record column specs, and narration templates — with the procedural bodies appropriately deferred to context files rather than inlined as pseudocode.	4 / 5
Workflow Clarity	A clearly sequenced four-step flow (settle objective → discover & score → compose/partition → land) with explicit validation gates, edge-report feedback loops (below_floor/above_ceiling, needs_anchor, no_base), a "Only on yes" compose gate, and a dedicated Failure modes section with halt-and-surface behavior.	5 / 5
Progressive Disclosure	Well-signaled one-level-deep `context/*.md` references with explicit read instructions and clearly organized sections, but the SKILL.md body is denser than an overview and the referenced context files cannot be verified against an actual bundle (no references/scripts/assets dirs are present).	4 / 5
	Total	16 / 20 Passed

Description

100%Weight 40%Scale 1-5

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

A strong, third-person description that concretely names the skill's actions, supplies natural trigger phrases, and cleanly answers both what and when while staking out a distinct niche. No verbosity or over-claim penalties apply.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — "discover the signals behind the brief, score them, and compose, partition, or cross them into a measured result" — covering the skill's full capability surface comprehensively.	5 / 5
Completeness	Explicitly answers both what ("Build a person audience from a plain-English description …") and when ("Use when a build starts from a description — …") with concrete trigger phrases.	5 / 5
Trigger Term Quality	Includes natural quoted triggers a user would actually say — "build me an audience of about 1–5M weekend hikers", "the highest-intent people in-market for solar", "where do these people cluster", "a lead list of in-market companies" — with good synonym/variation coverage.	5 / 5
Distinctiveness Conflict Risk	Carves a clear niche — "anchored on a description (no owned list)", "crossing the employment graph to their employers (or employees)", "behind /watt:audience" — and explicitly distinguishes from the sibling list-anchor skill, minimizing wrong-skill triggers.	5 / 5
	Total	20 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: wattdata/plugin
Path: skills/audience-generate-search/SKILL.md
Commit: 880e260

Reviewed: 4 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.