Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
the king is watching
v1.0.0AI Workflow Enforcer inspired by the Steam game "The King Is Watching". Just like subjects in the game only work when the King's gaze is upon them, this tool...
⭐ 1· 71·0 current·0 all-time
by@d0roro
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description (AI workflow enforcer: forced sequence, heartbeat, checkpointing, translation) align with the included code and docs. The repository contains a TaskTranslator that claims to convert NL→YAML→Python and then execute that code; that capability is powerful and consistent with the stated purpose, but is also a potential source of unexpected behavior if not constrained.
Instruction Scope
SKILL.md instructs runtime use of translate_and_run and producing enforced workflows. The docs and tests show the skill writes persistent state to .overseer state directories, runs background threads (heartbeats, reporter), and can run generated Python workflows. Those instructions do not explicitly constrain what generated code can do (file I/O, subprocesses, network). The test suite also inserts an absolute sys.path (/root/.openclaw/...) which is an odd hard-coded assumption.
Install Mechanism
No install spec is provided (instruction-only for the skill registry), but the package contains Python source and setup.py. No remote downloads, no extract-from-URL steps were specified in the metadata — low install mechanism risk. The package exposes a console entry point mapping to test_overseer:main which will run tests if installed as a script.
Credentials
The skill declares no required environment variables or credentials, which is proportional to basic offline workflow control. However the docs mention notification integrations (Feishu, Discord) and background notifications (notify_channel), which in practice will require API tokens or webhooks — those are not declared nor documented as required. If you configure notifications, you must supply credentials; absence from the manifest means the skill won't request them up-front, so manually inspect code that performs notifications before providing secrets.
Persistence & Privilege
The skill writes persistent state files (default .overseer, archive directories) and runs background threads to maintain heartbeats and periodic reports. That level of persistence and background activity is consistent with purpose, but it does mean the skill will create and modify files in your agent environment and can run asynchronous tasks. Because the TaskTranslator can generate and execute Python, persisted workflows could contain user-generated code that will be executed on your host.
What to consider before installing
What to check before installing or running this skill:
1) Inspect translator and runtime execution paths: open overseer/translator.py and overseer/__init__.py (the files that compile/execute translated workflows). Confirm whether the translator executes arbitrary Python, spawns subprocesses, or performs network calls. If it uses exec()/eval() or writes and imports generated .py files, assume arbitrary code execution is possible.
2) Run in an isolated environment first: test the skill in a disposable VM or container with minimal permissions and no sensitive mounted files. That limits damage if generated workflows perform unwanted actions.
3) Check notification integrations: search the codebase for Feishu/Discord/HTTP clients or webhook usage. If you plan to enable notifications, provide tokens only after auditing the implementation and using scoped secrets / webhooks rather than full-account credentials.
4) Review persistence paths and retention: the skill uses a .overseer state directory by default. Confirm where it writes files and whether those files might contain sensitive data. Consider configuring a dedicated state_dir on non-sensitive storage.
5) Be cautious with the one-liner/auto-run features: translate_and_run and auto-generated skill code can create and immediately execute workflows. Prefer the translate→preview→run pattern (translator.explain_plan, Overseer.from_plan) so you can inspect the generated plan before execution.
6) Note oddities in test code: test_overseer.py inserts an absolute path (/root/.openclaw/...) which may indicate assumptions about environment layout; that is not harmful by itself but worth being aware of.
If you cannot review the translator/runtime code yourself, treat the skill as untrusted: avoid running it with access to secrets, system-level directories, or production data.Like a lobster shell, security has layers — review code before you run it.
latestvk976djwa3tx1w14r6wn2geqqgd83qz8blatest execution-control long-running-tasks workflow task-automationvk976djwa3tx1w14r6wn2geqqgd83qz8b
License
MIT-0
Free to use, modify, and redistribute. No attribution required.