create-grant-update
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local
curlandpython3commands to fetch and process JSON data from the Karma protocol. These scripts are static and part of the skill's core functionality, used primarily for data formatting. - [DATA_EXFILTRATION]: The skill performs network requests to
gapapi.karmahq.xyzusing an API key provided via environment variables. This is the official API endpoint for the Karma protocol and is required for the skill's operation. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection via the data it processes from external sources.
- Ingestion points: External data is ingested from the Karma API via project and grant lookup endpoints.
- Boundary markers: Absent; the skill does not use specific delimiters to separate external content from instructions.
- Capability inventory: The skill uses
curlfor network operations andpython3for data processing and JSON parsing. - Sanitization: External data is parsed as JSON, but specific string fields (like project titles and grant names) are not sanitized for potential embedded instructions that could influence the LLM's behavior.
Audit Metadata