resync-upstream-skills
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill is designed to 'pull upstream skill releases' and resolve 'the latest stable release from the source.' It fetches content from external repositories or URLs defined in the local repository's attribution ledger (e.g.,
THIRD-PARTY.md). - [REMOTE_CODE_EXECUTION]: By ingesting external skill files—which represent the instructions the agent follows—there is a risk that malicious upstream content could influence the agent's behavior. Additionally, the skill instructions include running a 'repository's validator or skills CLI' on the downloaded content, which could be exploited if the content is malicious.
- [COMMAND_EXECUTION]: The skill requires the agent to execute shell commands for version control and validation, specifically 'Git history' lookups, 'committing and propagating' changes, and using a 'validator or skills CLI.'
- [INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection because it processes untrusted data from upstream sources.
- Ingestion points: External skill repositories, release notes, and commit history identified in
THIRD-PARTY.mdand the attribution ledger. - Boundary markers: The instructions do not specify any boundary markers or 'ignore' instructions for the content being synchronized.
- Capability inventory: The skill utilizes file system writes, Git commands, and local validation CLI tools.
- Sanitization: No sanitization or validation of the upstream content's instruction safety is mentioned beyond structural validation.
Audit Metadata