generate-local-issue
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a bundled script, scripts/optional-cli.mjs, to invoke the local issue-flow CLI via node:child_process.spawnSync. This integration is designed for read-only repository policy discovery and initialization checks. The execution is bounded by a five-second timeout to prevent resource exhaustion, and arguments are passed as a controlled array to mitigate shell injection vulnerabilities.
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from external sources such as problem documents, specifications, and repository configuration files (e.g., CONTRIBUTING.md).
- Ingestion points: Untrusted content enters the context during the resolution of local issue documents and GitHub issue descriptions as described in references/issue-input.md and references/issue-authoring.md.
- Boundary markers: While explicit technical delimiters are not used, the skill instructions in references/repository-policy.md provide strong directives to inspect all configuration as plain text and strictly avoid executing arbitrary configuration code.
- Capability inventory: The agent possesses the capability to write local issue artifacts (issue.md, metadata.json), perform Git repository operations, and execute the scoped issue-flow CLI.
- Sanitization: The scripts/conventions.mjs helper provides normalization and slugification for file and branch names, ensuring that untrusted title content is transformed into safe path segments.
Audit Metadata