opencode-session-inventory
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform local process and listener inspection using host-appropriate tools (e.g., to identify PIDs, ports, and working directories). It also utilizes the
herdrCLI for snapshotting workspace state and retrieving process information. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external local sources, including OpenCode API responses (
GET /api/session/active) and process inspection results. This data is then used to generate a report in a knowledge base or document store. Maliciously crafted session names or process metadata could potentially influence the agent's behavior during the reporting phase. - Ingestion points: Local process metadata (SKILL.md), OpenCode API endpoints (SKILL.md), and Herdr API snapshots (SKILL.md).
- Boundary markers: The skill does not explicitly define delimiters for external data, although it mandates the use of a table format for the final report.
- Capability inventory: The skill has the capability to read local process lists, query local network APIs, and write to a user-specified document store or knowledge base.
- Sanitization: The skill includes instructions to exclude prompts, message history, and credentials, but lacks specific sanitization rules for the metadata fields being reported.
Audit Metadata