infisical-pam

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill enables an agent to interact with and process data from external databases, servers, and cloud environments. This interaction creates a vulnerability surface where malicious instructions could be embedded in the data accessed via the PAM proxy.\n
  • Ingestion points: Infrastructure targets including PostgreSQL, MySQL, MongoDB, SSH servers, and Kubernetes clusters as described in references/accounts-and-access.md.\n
  • Boundary markers: No specific delimiters or instructions to ignore embedded commands within the brokered session data are defined.\n
  • Capability inventory: The agent can execute database queries and shell commands across the connected infrastructure.\n
  • Sanitization: The instructions do not specify methods for sanitizing or escaping the data retrieved from the brokered sessions.\n- [COMMAND_EXECUTION]: The skill details how to use the Infisical CLI to establish secure connections and sandbox agent processes.\n
  • Evidence: Documented use of infisical pam agentic access in SKILL.md and references/sessions-and-agents.md to broker access for AI agents.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 11:05 PM
Security Audit — agent-trust-hub — infisical-pam