Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
gpt-go
v1.0.1Enter GPT strong execution mode for the current conversation when explicitly invoked, such as with /gpt-go. Persist across the conversation. Default to execu...
⭐ 0· 197·0 current·0 all-time
byJoey-Chen@c-joey
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description, SKILL.md content, gateway prompt, and the small helper script all align: the skill's goal is to switch the agent into an aggressive execution mode and persist that behavior. The included sync-gateway.js simply extracts sections from SKILL.md to produce gateway-prompt.txt—this is coherent with the stated purpose.
Instruction Scope
The SKILL.md explicitly instructs the agent to 'Start executing when intent is clear', 'Inspect the environment directly before asking the user for inspectable facts', and to treat short directives as authorization to continue. Those instructions grant broad operational discretion and reduce confirmation steps. Even though pause boundaries are listed for high-risk actions (credentials, destructive changes, publishing, costs, production impact), the default-to-continue behavior can still cause the agent to perform many changes automatically unless deployment enforces stricter external constraints.
Install Mechanism
No install spec; this is instruction-only plus a small local helper script. sync-gateway.js only reads SKILL.md and writes gateway-prompt.txt—no external downloads, no archive extraction, and no network calls. Low install risk.
Credentials
The skill declares no required environment variables or credentials, but its instructions encourage directly inspecting the environment. That is a mismatch: the skill may attempt to read files, environment variables, or system state at runtime even though no such access is declared or justified. This makes the skill's runtime expectations disproportionate to its declared requirements.
Persistence & Privilege
The skill is user-invocable and not 'always: true' (good). Autonomous invocation (disable-model-invocation: false) is permitted by default, which is normal. However, combining autonomous invocation with the 'default to execute' instructions increases risk if the agent has tool access; there is no direct evidence the skill requests persistent system-level privileges or modifies other skills.
What to consider before installing
This skill changes the agent's behavior to act with minimal confirmation and to inspect the environment before asking. If you install it, do not simultaneously grant the agent unrestricted tool access (shell, filesystem, cloud credentials, or outgoing network) unless you trust it and run it in a safe sandbox. Prefer testing in an isolated environment first, enable strict logging/auditing of actions, and require explicit confirmations for destructive, credential-related, publishing, or production-impacting actions. If you need the behavior but worry about accidental access, consider modifying the SKILL.md to remove 'inspect environment directly' or to add explicit checks/confirmations for file/credential access.Like a lobster shell, security has layers — review code before you run it.
latestvk97aq389qchpjdh0t7dk28kabx8364a2
License
MIT-0
Free to use, modify, and redistribute. No attribution required.