update-issue-tree
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell commands and local bash scripts (
reassign-lib.sh,reassign-sub-issue.sh) to interact with the GitHub API via theghCLI. It manages environment-specific path resolution for these scripts across different possible installation directories to ensure portability.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from GitHub Issue titles and bodies to generate summaries and update the root issue. This presents an attack surface for indirect prompt injection if the ingested issue content contains instructions meant to manipulate the agent's behavior during report generation.\n - Ingestion points: GitHub Issue metadata and body content fetched via
gh apiin Step 1.\n - Boundary markers: No specific boundary markers (e.g., XML tags or clear delimiters) are implemented to isolate untrusted issue content within the generated issue bodies, increasing the risk that the agent may interpret issue content as instructions.\n
- Capability inventory: The skill possesses the capability to perform complex GitHub API operations (GET/POST/DELETE) and execute local shell scripts.\n
- Sanitization: The skill employs shell quoting for variable interpolation to prevent command injection, but it does not perform semantic sanitization of the issue text against adversarial prompt patterns.\n- [DYNAMIC_EXECUTION]: The skill dynamically resolves the path to its helper library by iterating through a list of candidate filesystem locations at runtime before sourcing the code. This ensures compatibility with different deployment structures but involves runtime code loading from paths determined during execution.
Audit Metadata