Rule Pack: owasp-agentic¶
OWASP LLM Top 10 agentic safety rule pack.
The scanning rules use matches, which is capped at 4096 characters: an
agent_input or agent_output whose content is longer fails the
evaluation closed with EvaluationError rather than going unscanned.
Split longer content into several facts of at most 4096 characters,
overlapping by the longest match you need to catch (an SSN or email
address straddling a boundary is otherwise missed).
Pack version: 1.0
Rule count: 4
Modules: owasp
Templates: agent_input, agent_output, tool_request
Rules¶
| Name | Salience | Action | Reason | Source |
|---|---|---|---|---|
detect-prompt-injection |
100 | escalate |
Possible prompt injection detected in agent input | src/fathom/rule_packs/owasp_agentic/rules/owasp_rules.yaml |
deny-excessive-agency-exec |
50 | deny |
Tool call is in the dangerous tools list (LLM04: Excessive Agency) | src/fathom/rule_packs/owasp_agentic/rules/owasp_rules.yaml |
flag-insecure-output-ssn |
90 | escalate |
Agent output may contain SSN pattern (LLM06: Insecure Output) | src/fathom/rule_packs/owasp_agentic/rules/owasp_rules.yaml |
flag-insecure-output-email |
80 | escalate |
Agent output may contain email address (LLM06: Insecure Output) | src/fathom/rule_packs/owasp_agentic/rules/owasp_rules.yaml |