keeper-admin

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill involves the installation of the official 'keepercommander' Python package from the public registry.
  • [COMMAND_EXECUTION]: Executes the 'keeper' and 'ksm' CLI tools to perform vault and enterprise operations, and uses 'tmux' for managing interactive session state and persistence.
  • [DATA_EXFILTRATION]: Provides capabilities to export vault records and audit reports to local files and includes instructions for running a REST API server.
  • [DYNAMIC_EXECUTION]: Includes patterns for generating and executing Python scripts at runtime via 'ksm exec' for data processing and reporting tasks.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from vault records which could potentially contain malicious instructions.
  • Ingestion points: Vault record titles, fields, and metadata retrieved via browse and search commands in SKILL.md.
  • Boundary markers: Absent; the skill does not define specific delimiters for untrusted vault content.
  • Capability inventory: File system writes (export), network hosting (api-server), and subprocess execution (keeper, ksm, tmux).
  • Sanitization: Relies on agent-level safety filters and explicit warnings to avoid printing sensitive secret fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 07:33 PM
Security Audit — agent-trust-hub — keeper-admin