company-brainify

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on the execution of numerous shell commands for filesystem manipulation, search, and validation, including 'mkdir', 'rm', 'cp', 'rsync', 'grep', 'wc', and 'find'.
  • [REMOTE_CODE_EXECUTION]: The workflow requires using 'git clone' and 'git push --force' with remote URLs, and explicitly depends on the external 'git filter-repo' tool to perform destructive history rewriting.
  • [DATA_EXFILTRATION]: The skill's primary purpose is moving data from a personal environment to shared remote repositories. This inherent behavior could be misused to exfiltrate sensitive data if the destination URL is replaced with an attacker-controlled endpoint.
  • [INDIRECT_PROMPT_INJECTION]: The skill is highly exposed to indirect prompt injection as it processes untrusted data. 1. Ingestion points: The skill reads markdown content from 'people/', 'meetings/', and 'daily/' directories, which are likely to contain inputs from various sources. 2. Boundary markers: There are no explicit delimiters or 'ignore' instructions surrounding the content being processed to prevent the agent from being subverted by embedded commands. 3. Capability inventory: The agent possesses full shell access, filesystem read/write, and network access via Git. 4. Sanitization: The logic depends on the agent reading and judging content for sensitivity, which requires the agent to fully process potentially malicious instructions hidden in the data.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 06:26 AM
Security Audit — agent-trust-hub — company-brainify