cloudflare-access-add-user

Warn

Audited by Gen Agent Trust Hub on Feb 24, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill accesses and modifies the /home/dawiddutoit/projects/network/.env file, which contains sensitive administrative credentials such as CLOUDFLARE_ACCESS_API_TOKEN and CLOUDFLARE_ACCOUNT_ID. Accessing these secrets directly exposes them to the agent context.\n- [COMMAND_EXECUTION]: The agent is instructed to execute a local shell script (/home/dawiddutoit/projects/network/scripts/update-access-emails.sh) using the Bash tool to apply policy changes via the Cloudflare API.\n- [DYNAMIC_EXECUTION]: The skill requires the agent to dynamically modify the contents of the update-access-emails.sh script by injecting user-provided email addresses into a JSON array within the script file prior to execution, which is a significant security concern regarding runtime code modification.\n- [INDIRECT_PROMPT_INJECTION]: The skill interpolates untrusted user input (email addresses) into configuration and script files, creating a potential attack surface.\n
  • Ingestion points: User-provided email strings as defined in the SKILL.md instructions.\n
  • Boundary markers: No explicit delimiters or boundary markers are used in the script modification step to isolate user input.\n
  • Capability inventory: The skill possesses 'Edit' (file modification) and 'Bash' (command execution) capabilities, allowing it to apply changes and run scripts.\n
  • Sanitization: A Python regex-based validate_email function is suggested to the agent for validating input before processing, which provides some mitigation against malformed input.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 24, 2026, 05:23 PM