opencode-agent

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands using user-provided instructions and directory paths, specifically via the opencode run command.
  • [DATA_EXFILTRATION]: The skill accesses session metadata in ~/.copilot/session-state and transfers files from the filesystem to a workspace directory for transmission via Telegram.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and parsing data from external session logs. Ingestion points: Reads JSON data from /tmp/session.json and events.jsonl from the Copilot session directory. Boundary markers: External data is processed and displayed without isolation delimiters or instructions to ignore embedded commands. Capability inventory: The skill has access to shell execution via CLI tools and file system manipulation. Sanitization: No sanitization or validation is applied to the ingested session content before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 02:57 AM