memory-connect

Warn

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of a global NPM package @mitosislabs/sdk from the public registry. This allows the execution of code downloaded from an external source on the user's machine.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands (mi login, mi cortex manifest, mi connect). These commands are used to perform authentication, retrieve remote manifests, and modify local configuration files to persist the connection.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads external data provided by the mi cortex manifest command, which could potentially contain untrusted content.
  • Ingestion points: Command output of mi cortex manifest as described in SKILL.md.
  • Boundary markers: Absent. There are no instructions provided to the agent to treat this output as untrusted or to use delimiters.
  • Capability inventory: The skill has the capability to execute shell commands and write to the local file system (via mi connect).
  • Sanitization: Absent. The agent is instructed to "see what it holds" without explicit validation or filtering steps.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 27, 2026, 06:18 AM
Security Audit — agent-trust-hub — memory-connect