darkweb-intel

Warn

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: MEDIUMPERSISTENCEPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [PERSISTENCE]: The skill uses crontab to schedule automated monitoring of external sources. It provides instructions to set up recurring daily sweeps of Ahmia search results and IntelX intelligence feeds to track target terms over time.
  • Evidence: SKILL.md and guides/dark-web-investigation-playbook.md include command patterns like echo "0 8 * * * curl ..." | crontab - to establish periodic data collection.
  • [PRIVILEGE_ESCALATION]: The skill requires administrative access via sudo for environment provisioning and service management. This includes installing network tools (Tor, torsocks), managing system-level daemons, and modifying network configuration files in /etc/.
  • Evidence: payloads.md and SKILL.md document setup steps such as sudo apt install tor, sudo service tor start, and editing /etc/tor/torrc.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process data from untrusted dark web sources, including marketplaces and threat actor forums. There are no boundary markers or content sanitization steps to prevent the agent from following instructions hidden within the scraped content.
  • Ingestion points: Python-based scraping logic and curl commands targeting .onion search results and forum profiles.
  • Boundary markers: Not present; the skill lacks delimiters or warnings for the agent to ignore embedded instructions in scraped data.
  • Capability inventory: The skill possesses extensive capabilities via Bash, WebFetch, and Write tools that could be abused if an injection occurs.
  • Sanitization: No filtering or sanitization is applied to the raw HTML or JSON data retrieved from external sources.
  • [COMMAND_EXECUTION]: The skill employs powerful system commands for data management and environment setup. This includes the use of shred for secure data deletion to prevent forensic recovery and dd for overwriting block devices when creating bootable Tails media.
  • Evidence: payloads.md contains shred -uvz commands and a sudo dd command for imaging a USB drive.
  • [DATA_EXFILTRATION]: As part of its OPSEC and counter-OSINT workflows, the skill audits and manipulates sensitive user data, including SSH keys, GPG keys, and shell history. While intended for security auditing, these patterns involve accessing sensitive local paths.
  • Evidence: guides/dark-web-investigation-playbook.md and payloads.md instruct the agent to inspect ~/.ssh, ~/.gnupg, and shred ~/.bash_history.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 19, 2026, 04:36 PM