manage-agent-identity
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process session traces, audit logs, and identity registration records which are derived from active agent sessions and may contain untrusted data.
- Ingestion points: The 'Active Session Monitoring' and 'Audit & Compliance' sections of
SKILL.mddescribe the ingestion of trace spans, tool call logs, and resource access records into the audit system. - Boundary markers: The instructions lack explicit text-based delimiters (such as XML tags or specific markdown blocks) to isolate external trace data from the agent's core instructions.
- Capability inventory: The skill has access to privileged tools including
execute_command,write_file,edit_file, andcreate_fileas defined in theallowed-toolsfrontmatter ofSKILL.md. - Sanitization: The skill explicitly addresses this risk in section 'ASI06 Memory & Context Poisoning' by stating that traces are untrusted inputs and must be validated and redacted before archival, which serves as a mitigation measure.
Audit Metadata