quickgo-skill
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [Arbitrary File Write]: The
scripts/rest_request.pyscript includes a functionality to save raw request output to the local file system. By setting thesave_rawparameter to true, the agent can write the content of a network response to a path specified byraw_output_path. This provides a mechanism for the skill to create or modify files on the local environment using data retrieved from remote sources. - [Generalized Network Access]: The utility script
scripts/rest_request.pyis designed as a generic REST client. Although the instructions inSKILL.mddirect the agent to use the European Bioinformatics Institute's QuickGO API, the script itself is capable of performingGETandPOSTrequests to any URL provided in the input payload. This generalized capability could be leveraged to interact with other network-accessible resources. - [Indirect Prompt Injection Surface]: The skill is designed to ingest and summarize data from external API endpoints, which introduces a potential surface for indirect prompt injection. If an external service returns content containing malicious instructions, the agent might inadvertently act upon them during its summarization or subsequent processing steps.
- Ingestion points: Data is retrieved from external URLs within the
executefunction ofscripts/rest_request.pyusing therequestslibrary. - Boundary markers: The skill instructions do not currently define specific delimiters or instructions to ignore potential commands embedded in the retrieved data.
- Capability inventory: The skill has the capability to perform network requests and write files to the local disk.
- Sanitization: The script implements data truncation for display purposes but does not perform validation or sanitization of the content fetched from the remote API.
Audit Metadata