upstream-patches
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute git commands and a local script
./scripts/upstream.shto manage the patch stack and repository state. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external patch files which could contain hidden instructions targeting the agent.
- Ingestion points: The agent reads
patches/*.patchusingrg,grep, andsedto find and inspect patches. - Boundary markers: There are no instructions to the agent to treat patch content as untrusted or to use delimiters.
- Capability inventory: The agent is granted capabilities to delete files (
rm), modify git history, and run a local management script. - Sanitization: No sanitization or validation of the ingested patch content is performed.
Audit Metadata