cli-contexts
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions direct the installation of the 'omniroute' package from the NPM registry using 'npm install -g omniroute' or execution via 'npx'.
- [COMMAND_EXECUTION]: The skill operates by executing the 'omniroute' CLI tool and its various subcommands to manage context configurations, RTK filters, and conversation sessions.
- [DATA_EXFILTRATION]: The skill documents the 'contexts export' command, which includes an '--include-secrets' flag. This functionality allows the extraction of sensitive credentials (API keys and access tokens) to a local JSON file, which could be exploited by an agent to exfiltrate secrets if output is directed to a location accessible to the agent or an attacker.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection through the 'contexts import ' command, which processes external data files.
- Ingestion points: The 'contexts import ' command in SKILL.md allows the agent to ingest and apply configurations from external JSON files.
- Boundary markers: There are no documented boundary markers or specific warnings to ignore embedded instructions within the imported context files.
- Capability inventory: The skill has the capability to execute shell commands, manage session states, and export data to the file system.
- Sanitization: The skill documentation does not mention any validation or sanitization of the imported JSON content before it is processed by the CLI tool.
Audit Metadata