fabric-cli

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the official installation script from the vendor's domain (https://fabric.so/cli/install.sh). This is documented as a standard setup step, and the skill includes instructions for the agent to recommend manual review before execution.\n- [COMMAND_EXECUTION]: Interacts with the local environment via the fabric CLI. Bundled scripts use shlex.quote to sanitize user-provided strings and prevent command injection, ensuring that inputs are handled safely.\n- [DATA_EXFILTRATION]: Implements proactive security measures by including regex-based redaction in its helper scripts. This logic automatically removes API keys, tokens, and private keys from command outputs and generated notes to prevent accidental exposure.\n- [PROMPT_INJECTION]: \n
  • Ingestion points: Data retrieved from the Fabric library via search and browse operations enters the agent's context (SKILL.md, references/workflows.md).\n
  • Boundary markers: Missing specific prompt delimiters for external content, though the skill mitigates risk through automated redaction logic.\n
  • Capability inventory: File uploads, task management, and resource creation via the fabric CLI (SKILL.md, scripts/fabric_command_plan.py).\n
  • Sanitization: Implements automated secret redaction and shell-safe quoting for all processed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 05:28 PM
Security Audit — agent-trust-hub — fabric-cli