debugging-executions
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to perform debugging actions "unprompted" and not to ask the user for permission. This reduces human oversight and modifies the agent's default safety behavior regarding tool confirmation.
- [DATA_EXFILTRATION]: The skill facilitates access to resolved parameters from workflow executions. While it indicates that sensitive variables like
$secretsare not reconstructed during replay, this process still brings execution data into the agent's context. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by ingesting untrusted workflow execution data. Ingestion points: Workflow outputs and resolved parameters accessed via the
executionstool inSKILL.md. Boundary markers: Absent; there are no instructions to ignore or delimit embedded instructions within the processed data. Capability inventory: The agent utilizesexecutionsandworkflowstools. Sanitization: Absent; no validation or filtering is applied to the execution data.
Audit Metadata