improve-sdk-naming
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [Indirect Prompt Injection] (HIGH): The skill processes untrusted OpenAPI specifications through the
speakeasy suggestcommand. This creates an attack surface where an attacker can embed malicious instructions within the spec (e.g., in descriptions or operation IDs) to influence the AI's naming suggestions or downstream agent actions.\n - Ingestion points: OpenAPI spec file path provided via the
-sflag inSKILL.md.\n - Boundary markers: Absent; there are no instructions to the agent to ignore embedded commands within the processed data.\n
- Capability inventory: Execution of
speakeasy suggestcommands and file writing via the-ooutput flag.\n - Sanitization: Absent; the skill does not perform validation or sanitization of the external OpenAPI content before processing.\n- [Command Execution] (MEDIUM): The skill relies on executing the
speakeasyCLI. While the commands are specific, they depend on an external binary and environment state, which could be exploited if the CLI tool or the input spec is compromised.\n- [Credentials Unsafe] (LOW): The skill instructions involve setting theSPEAKEASY_API_KEYenvironment variable. While it uses placeholders, the workflow encourages the management of sensitive credentials in the shell environment, which is a common point of exposure.
Recommendations
- AI detected serious security threats
Audit Metadata