octocode-research
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to execute specific shell commands to initialize and manage a background daemon (server). This includes using
npm startfor spawning the detached process andlsof/killfor lifecycle management as detailed inSKILL.mdandsrc/server-init.ts. - [EXTERNAL_DOWNLOADS]: The skill fetches source code, metadata, and repository structures from external sources including GitHub, npm, and PyPI to facilitate its research functions.
- [DATA_EXFILTRATION]: The server implements a de-identified session telemetry system that tracks usage metrics and error rates. While it includes an opt-out mechanism (
LOG=false) and redacts sensitive keys from logs, it still transmits diagnostic data to the author's infrastructure. - [PROMPT_INJECTION]: The skill has a large indirect prompt injection surface as it ingests untrusted code from external repositories. However, it includes defensive instructions in
references/GUARDRAILS.mddirecting the agent to treat external code as display-only research data and to ignore any embedded commands or instructions in comments.
Audit Metadata