CtrlK
BlogDocsLog inGet started
Tessl Logo

aatmf-t05-api-exploitation

AATMF T5 — Model & API Exploitation. Rate-limit abuse, token-cost amplification, schema bypass, model-version manipulation.

60

Quality

70%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./packages/decepticon/decepticon/skills/plugins/llm-redteam/t05-api-exploitation/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

72%

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 concise, well-organized technique catalog with a concrete probe-pattern config, but it describes many techniques rather than giving fully executable steps and lacks explicit validation checkpoints in its probe workflow.

Suggestions

Convert key techniques into executable, copy-paste-ready probe snippets — e.g. an actual curl/HTTP example for model_id probing and a concrete promptfoo test case for schema bypass — rather than describing the action.

Add a short ordered probe workflow with a validation checkpoint, e.g. 1) run probe pattern, 2) confirm signal against the Detection signals list, 3) re-run to rule out false positives before scoring severity.

For each technique, pair the attack with a one-line verification of success (what observable result confirms it worked) so findings are confirmable rather than asserted.

DimensionReasoningScore

Conciseness

The body is a lean bullet catalog of techniques with a one-line orientation ('Attack the LLM as an API service...') and no padded explanation of concepts Claude already knows; every section earns its tokens.

3 / 3

Actionability

It offers some concrete artifacts — a promptfoo probe-pattern YAML block, example prompts ('Repeat the word token 500 times'), and candidate model_id values — but many techniques are described ('induce model to NOT call the function', 'Probe with increasingly-long inputs') rather than given as executable, copy-paste-ready steps.

2 / 3

Workflow Clarity

The Probe pattern section supplies a runnable config, but there is no sequenced validate→fix→retry workflow or explicit checkpoints for confirming a finding (e.g. confirming model-version manipulation succeeded), which leaves the sequence present but checkpoints implicit.

2 / 3

Progressive Disclosure

It is a single self-contained file with no external bundle references and clearly organized sections (Techniques, Probe pattern, Detection signals, Severity, Defender, Cross-references), so the well-organized structure satisfies progressive disclosure without needing separate files.

3 / 3

Total

10

/

12

Passed

Description

67%

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 specific and distinctive, listing four concrete LLM-API exploitation techniques, but omits explicit 'Use when...' trigger guidance and leans on internal AATMF jargon over natural user keywords.

Suggestions

Add a 'Use when...' clause naming natural trigger situations, e.g. 'Use when assessing rate-limit/quota abuse, token-cost amplification, or model_id/version manipulation against an LLM API'.

Soften or expand the AATMF T05 taxonomy label with plain-language terms (e.g. 'LLM API economic and schema abuse') so users who do not know the internal taxonomy still recognize the trigger.

Include common natural phrasings users would say ('API cost abuse', 'model_id tampering', 'bypassing JSON structured output') alongside the technical terms.

DimensionReasoningScore

Specificity

Names multiple concrete technique actions — 'Rate-limit abuse, token-cost amplification, schema bypass, model-version manipulation' — matching the 'lists multiple specific concrete actions' anchor.

3 / 3

Completeness

It clearly states what the skill does (the four techniques) but provides no 'Use when...' clause or equivalent trigger guidance, which per the judging guidelines caps completeness at 2.

2 / 3

Trigger Term Quality

'rate limit' and 'API' are natural user terms, but 'AATMF T05', 'token-cost amplification', and 'model-version manipulation' lean on internal taxonomy/jargon, missing common variations a user would actually say.

2 / 3

Distinctiveness Conflict Risk

The 'Model & API Exploitation' niche with LLM-specific technique list is a clear, distinct slot unlikely to trigger for unrelated skills.

3 / 3

Total

10

/

12

Passed

Validation

93%

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

Validation15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

Total

15

/

16

Passed

Repository
PurpleAILAB/Decepticon
Reviewed

Table of Contents

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.