llm-redteam-overview

LLM red team category — full AATMF v3 tactic coverage (T01–T15). Routing skill: read this first to identify which tactic applies, then load the matching sub-skill. Maps to MITRE ATLAS where overlap exists.

Quality

71%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./packages/decepticon/decepticon/skills/plugins/llm-redteam/SKILL.md

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 tight, well-structured routing catalog with clear navigation and lean token usage. Its main weakness is the Tooling section, which describes tools rather than providing executable commands or invocation examples.

Suggestions

Turn the Tooling table entries into actionable invocations (e.g., a sample `promptfoo` / `garak` / `pyrit` command line) so guidance is copy-paste ready rather than descriptive.

Add a one-line note on how to verify a chosen tactic matched (e.g., confirm the target surface before loading the sub-skill) to strengthen the routing workflow's validation checkpoint.

Dimension	Reasoning	Score
Conciseness	The body is lean and dense with pointers — a tactic map table, a routing tree, a compact tooling table, and brief cross-references — with no padding or re-explanation of concepts Claude already knows.	3 / 3
Actionability	Provides concrete `load_skill(...)` load paths and a decision tree, but the Tooling table describes each tool's purpose rather than giving executable commands or examples, leaving the guidance instructive rather than copy-paste ready.	2 / 3
Workflow Clarity	The routing process is clearly sequenced (identify target type -> match tactic -> load sub-skill) with a quick-routing decision tree and a 'When in doubt' fallback (T01 + T05 probe); no destructive/batch operations are present that would require validation checkpoints.	3 / 3
Progressive Disclosure	Acts as a well-organized catalog overview with explicit one-level-deep references (per-tactic `load_skill` load paths) and clear navigation via the tactic map and quick-routing tree; no monolithic wall or nested references.	3 / 3
	Total	11 / 12 Passed

Description

57%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 clearly conveys the skill's niche and routing role but leans on framework jargon (AATMF, MITRE ATLAS) and lacks an explicit 'Use when...' trigger clause, which caps completeness. Natural trigger terms are present but under-represented relative to the technical vocabulary.

Suggestions

Add an explicit 'Use when ...' clause naming natural user triggers (e.g., jailbreaks, prompt injection, AI red teaming) directly in the description so the 'when' is not merely implied.

Reduce reliance on framework abbreviations (AATMF v3, T01–T15, MITRE ATLAS) in favor of plain-language terms a user would actually say, keeping the trigger_term_quality score's natural vocabulary.

Briefly state the concrete actions the skill itself performs beyond routing (e.g., 'runs red-team probes and maps findings to tactics') to lift specificity.

Dimension	Reasoning	Score
Specificity	Names the domain ("LLM red team category") and concrete routing actions ("identify which tactic applies", "load the matching sub-skill", "Maps to MITRE ATLAS"), but these are routing actions rather than a comprehensive list of the skill's own concrete capabilities.	2 / 3
Completeness	Clearly answers "what" (a routing skill that identifies the applicable tactic and loads a sub-skill) but the "when" is only implied via "read this first"; per the guideline, a missing explicit 'Use when...' clause caps completeness at 2.	2 / 3
Trigger Term Quality	Contains some natural terms users would say ("LLM red team", "prompt injection") but is dominated by framework jargon ("AATMF v3", "T01–T15", "MITRE ATLAS") and misses common variations; the richer natural triggers live in metadata.when_to_use rather than the description.	2 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (AI/LLM red-teaming) with distinct triggers unlikely to conflict with general-purpose skills.	3 / 3
	Total	9 / 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
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: PurpleAILAB/Decepticon
Path: packages/decepticon/decepticon/skills/plugins/llm-redteam/SKILL.md
Commit: 4484f85

Reviewed: 10 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.