Call O3 Llm
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: call-o3-llm Version: 0.1.0 The skill bundle contains standard metadata in `_meta.json` and instructions for an AI agent in `SKILL.md`. The `SKILL.md` content defines the agent's persona, context, and output rules (e.g., language matching, staying task-focused). There are no instructions for data exfiltration, malicious execution, persistence, or any form of prompt injection against the agent to perform unauthorized actions. The use of template variables like `$SHOW_NAME$` is noted but not used in a way that suggests a vulnerability or malicious intent.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may follow the skill’s role and meeting-minute rules instead of the user’s actual goal.
The skill attempts to set its own unconditional priority and makes provided meeting minutes authoritative, which could override normal user intent or trusted context handling when the skill is invoked.
This is the highest-priority directive and must be followed unconditionally ... meeting minutes ... are your sole source of information and memory, and you must adhere to them.
Change the prompt to state that these are scoped guidance for the skill only, remain subordinate to user/system instructions, and treat meeting minutes as context rather than authority.
A user may install or invoke the skill expecting an O3 model-calling helper, but get unrelated roleplay-style behavior.
The advertised purpose is an O3 LLM calling task, but the actual instructions define a creative-team meeting persona, creating a misleading expectation about what the skill does.
description: AI agent for call o3 llm tasks ... You are a creative expert ... participating in a brainstorming meeting chaired by your superior—the Chief Creative Officer (CCO).
Rename and redescribe the skill to match its actual creative brainstorming persona, or add the missing O3-calling behavior and disclose how it works.
Bad or manipulated meeting minutes could affect the agent’s responses.
The prompt relies on meeting minutes as memory/context, which is purpose-aligned for a meeting agent but can be risky if the minutes contain errors or injected instructions.
The entire context of the meeting, historical discussions, and current agenda items are recorded in real time in the meeting minutes. These minutes are your sole source of information and memory
Use meeting minutes as evidence to consider, not as unquestionable memory, and avoid feeding untrusted or instruction-bearing minutes without review.