Caveat verdict
agent-reach
agent-reach-skip
This skill provides read/search/interact access to 13+ public platforms (Twitter, Reddit, YouTube, GitHub, etc.) using documented CLI tools and public APIs, all consistent with its stated purpose as an internet-access layer for agents.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Permission integrity
network_out
Findings (8)
Possible hardcoded credential
agent_reach/guides/setup-twitter.md · code · TOKEN="你的auth_token
Pipe-to-shell pattern (curl | sh) — supply chain attack vector
agent_reach/cli.py · prose · downgraded · curl NodeSource setup | bash
Pipe to bash — executes piped content as shell commands
agent_reach/cli.py · prose · downgraded · | bash
subprocess execution — runs system commands from Python
agent_reach/channels/bosszhipin.py · prose · downgraded · subprocess.run(
Downloads executables from external URLs
agent_reach/cli.py · prose · downgraded · download from https://
Python os.environ.get — reads environment variable
agent_reach/channels/bilibili.py · prose · downgraded · os.environ.get(
importlib.import_module — dynamic module loading
agent_reach/cli.py · prose · downgraded · importlib.import_module(
Python shutil file operation — copies/moves/deletes files
agent_reach/cli.py · prose · downgraded · shutil.rmtree(
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class A/D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
network_out Is this flag fair?
Thanks — recorded.