omh-provider-profile-posture
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute specific command-line operations using the
omhutility (e.g.,omh runtime record,omh ops provider-profile-posture). These commands are standard for the workflow defined by the author ('oh-my-hermes maintainers') and are used to track status and process metadata. - [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for potentially untrusted data through the processing of JSON input files.
- Ingestion points: The command
omh ops provider-profile-posture --input <file>reads data from an external file specified by the user or environment. - Boundary markers: The instructions specify the use of "not-evidence boundaries" and require the agent to separate prepared guidance from observed platform/runtime evidence.
- Capability inventory: The skill environment permits file reading and execution of the
omhCLI tool. - Sanitization: The skill mandates strict schema validation (version
provider_profile_posture_input/v1) for the input JSON, which limits the interpreter's exposure to arbitrary fields, although it does not fully eliminate the risk of malicious natural language instructions embedded within those fields.
Audit Metadata