github-project-management
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external sources, which can be manipulated to influence agent behavior.
- Ingestion points: Fetches issue bodies, titles, and comments using
gh issue viewandgh issue list(found in multiple examples inSKILL.md). - Boundary markers: Absent. External content is interpolated directly into shell variables and agent prompts without delimiters or instructions to ignore embedded commands.
- Capability inventory: The skill has extensive capabilities, including full GitHub CLI access (
gh), creating/editing issues, and executing arbitrary commands vianpxand shell scripts. - Sanitization: None. There is no evidence of validation or escaping for the data retrieved from GitHub before it is passed to other tools or agents.
- [COMMAND_EXECUTION]: The skill frequently uses shell scripts that incorporate variables derived from external GitHub data (e.g.,
ISSUE_BODY,TITLE,BODY). Maliciously crafted issue content could potentially lead to command injection if not handled carefully by the underlying shell environment. - [EXTERNAL_DOWNLOADS]: The skill instructions promote the execution of remote packages via
npx(claude-flow@alpha,ruv-swarm). These resources are identified as vendor-owned tools within the 'ruvnet' ecosystem and are part of the intended primary purpose.
Audit Metadata