omni
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download binary releases and checksum files from the vendor's repository at
github.com/fajarhide/omni. - [COMMAND_EXECUTION]: The skill provides instructions for system installation using
brew install,curl, andtar. It also defines operational commands likeomni initandomni doctor. An external installation script is mentioned atomni.weekndlabs.com/installbut the skill explicitly advises against its use for security reasons. - [INDIRECT_PROMPT_INJECTION]: OMNI modifies tool output by injecting markers (e.g.,
[OMNI: ... redacted]). This represents a surface for indirect prompt injection where malicious data from commands could attempt to mimic these markers to deceive the agent. - Ingestion points: Shell command output in the agent's context.
- Boundary markers: Uses specific bracketed syntax but lacks cryptographic or structural separation from untrusted output.
- Capability inventory: Includes shell execution and file retrieval via
omni retrieve. - Sanitization: Implements automated redaction of sensitive credential-like values.
Audit Metadata