Skip to content

AAK-LLM-002: LLM analysis did not cover a tool

Severity
INFO
Category
Tool Poisoning (tool-poisoning)
Scanner
No scanner module. The rule id is named in code by agent_audit_kit/models.py.
Auto-fixable
No.
SARIF name
LlmAnalysisIncomplete

What it detects

scan --llm-scan was asked to analyze a tool's metadata and could not: the provider credential is missing, the request failed, the reply was cut off before the verdict, the provider declined to analyze it, or the text exceeded the analysed length. The scan cannot show a clean result for what it skipped, so, like a crashed scanner, this makes the run INCOMPLETE.

How to fix

Fix the cause the evidence names and re-run, or pass --allow-scanner-failure to accept the gap. When the model declined to analyze a tool, read that tool's metadata yourself: a refusal is not a clean result.

To confirm the finding is gone, report only this rule:

agent-audit-kit scan . --rules AAK-LLM-002

Limitations

None stated. The registry records no limitations for this rule; that is not a claim that it has none.

References

No external references are recorded for this rule.


All rules. This page is generated from the rule registry when the documentation is built; to change it, change the rule in agent_audit_kit/rules/builtin.py. To give the rule a hand-written write-up, add docs/rules/AAK-LLM-002.md: the page then shows that text, with the registry reference appended under it.