agent-harness-selftest

Prueba de vida de los frenos de un arnes agentico: por cada regla verificable inyecta la violacion en un sandbox y exige que el freno muerda (audit en rojo o gate exit distinto de cero), con caso negativo obligatorio (el arnes conforme pasa) y veredicto tri-estado MUERDE/NO-MUERDE/OMITIDA. Usar al pedir 'probar que los frenos muerden', 'selftest del arnes', 'harness selftest', 'el gate esta vivo o muerto', 'regla sin comando que la haga fallar'. NO para puntuar calidad (agent-harness-audit) ni reparar (agent-harness-repair).

Quality

87%

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

75%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 well-structured, mostly lean body that gives a concrete runnable command, a sequenced workflow with validation and a repair feedback loop, and one-level references that resolve to real files. Minor gaps: a placeholder in the main command, a dense opening paragraph, and a Packet block advertising bundle layers that do not all exist.

Suggestions

Tighten the "Que es" paragraph into shorter sentences or a short list so the enumeration of brake types reads as efficiently as the rest of the body.

Make the primary command copy-paste ready by showing a concrete example directory in addition to the <dir-del-arnes> placeholder.

Reconcile the Packet block's advertised layers (knowledge/, prompts/, examples/, agents/) with the actual bundle, or note which layers are optional/unused so the structure is not over-claimed.

Dimension	Reasoning	Score
Conciseness	Lean, information-dense Spanish with no over-explanation of concepts Claude already knows, though the single "Que es" paragraph crams several brake types into one long sentence and could be tightened slightly. It is efficient rather than padded.	4 / 5
Actionability	Provides a concrete executable command ("scripts/selftest_harness.py <dir-del-arnes> --json") and specific verdict interpretation, but the command carries a placeholder argument so it is not fully copy-paste ready.	4 / 5
Workflow Clarity	A clear 5-step sequence with explicit validation checkpoints ("Cero probes corridas = rojo siempre; OMITIDA nunca cuenta como verde") and a feedback loop (NO-MUERDE -> rutear a reparar el FRENO); validation is present so the destructive/batch cap does not apply, but a couple of checkpoints could be more explicit.	4 / 5
Progressive Disclosure	SKILL.md acts as an overview with a real one-level-deep reference (references/probes-y-limites.md, verified present) and real scripts; good section structure, though the Packet block advertises layers (knowledge/, prompts/, examples/, agents/) not present in the actual bundle.	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 states capabilities, supplies natural bilingual trigger phrases, explicitly defines when to use it, and draws clear boundaries against related skills. No first/second-person voice or vague fluff is present.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions with comprehensive coverage: "por cada regla verificable inyecta la violacion en un sandbox y exige que el freno muerda", mandatory negative case ("caso negativo obligatorio"), and tri-state verdict ("MUERDE/NO-MUERDE/OMITIDA").	5 / 5
Completeness	Explicitly answers both what (life-test of harness brakes via violation injection + bite enforcement + negative case + tri-state verdict) and when ("Usar al pedir ..." with concrete trigger phrases).	5 / 5
Trigger Term Quality	Comprehensive coverage of natural trigger phrases including synonyms in both Spanish and English: "probar que los frenos muerden", "selftest del arnes", "harness selftest", "el gate esta vivo o muerto", "regla sin comando que la haga fallar".	5 / 5
Distinctiveness Conflict Risk	Clear niche (harness brake selftest) with an explicit boundary clause — "NO para puntuar calidad (agent-harness-audit) ni reparar (agent-harness-repair)" — minimizing conflict with sibling skills.	5 / 5
	Total	20 / 20 Passed

Validation

68%

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 — 11 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
referenced_paths_exist	Referenced path issues: 1 missing	Warning

	Total	11 / 16 Passed

Repository: JaviMontano/claude-plugins
Path: plugins/claude-native-toolkit/skills/agent-harness-selftest/SKILL.md
Commit: e8f986b

Reviewed: 3 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.