bump-version
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads from project-level JSON files which may contain untrusted data if the repository has been tampered with.
- Ingestion points: Current version strings are read from
package.json,.claude-plugin/plugin.json, and.sd0x/install-state.jsonas specified inSKILL.md. - Boundary markers: The instructions do not define delimiters or specific constraints to prevent the agent from being influenced by other content in these files.
- Capability inventory: The skill leverages the agent's Edit tool to perform file write operations based on the data it reads.
- Sanitization: The skill does not specify validation or sanitization logic (e.g., verifying the version matches a SemVer regex) before the agent processes and propagates the values.
Audit Metadata