Skills
skills/roasbeef/claude-files/lnd/Gen Agent Trust Hub

lnd

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHCOMMAND_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADS
Full Analysis
  • COMMAND_EXECUTION (HIGH): The script scripts/docker-start.sh is vulnerable to shell command injection. It collects unhandled positional arguments into a variable that is later executed using eval without any sanitization or quoting.
  • Evidence: In scripts/docker-start.sh, lines 98-101 append arguments to COMPOSE_EXTRA_ARGS, and line 147 executes the final command via eval "$CMD".
  • CREDENTIALS_UNSAFE (HIGH): The skill contains multiple instances of hardcoded RPC credentials ('devuser' and 'devpass') across various scripts and Docker Compose templates. While intended for local test networks (regtest/simnet), they represent a hardcoded secret pattern.
  • Evidence: Found in scripts/bitcoin-cli.sh, scripts/mine.sh, scripts/multi-node-setup.sh, scripts/regtest-setup.sh, templates/docker-compose-multi.yml, templates/docker-compose.yml, and templates/start-lnd.sh.
  • EXTERNAL_DOWNLOADS (LOW): The build scripts retrieve source code and PR data from the lightningnetwork/lnd GitHub repository. This repository is not included in the provided list of trusted GitHub organizations.
  • Evidence: scripts/docker-build.sh and scripts/build-pr.sh use git fetch and gh pr checkout to pull remote code for image building.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 06:23 PM