← Back to Skills Marketplace
48
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install output-wrong-task
Description
The model produces correct-looking output that addresses a different task than the one requested — typically a related but distinct interpretation of an ambi...
Usage Guidance
This skill is low-risk: it's just text guidance for clarifying prompts and verifying outputs. Because it's instruction-only, it won't install or access system resources, but remember that when you invoke it the agent may request or operate on any user-provided sensitive content you include in prompts — avoid testing with secrets. If you want tighter control, only run it manually (user-invocable) and review outputs before acting on them.
Capability Analysis
Type: OpenClaw Skill
Name: output-wrong-task
Version: 1.0.0
The skill bundle consists solely of documentation and metadata (SKILL.md, _meta.json) describing a common LLM failure mode where a model misinterprets a prompt. It contains no executable code, no network requests, and no instructions that would cause the agent to perform unauthorized or harmful actions.
Capability Assessment
Purpose & Capability
The name and description (detecting and avoiding 'output wrong task') match the SKILL.md content: guidance for clarifying prompts and checking outputs. It requests no binaries, credentials, or config paths that would be unrelated to this purpose.
Instruction Scope
SKILL.md contains only guidance on how to restate, decompose, and verify prompts and outputs. It does not instruct the agent to read files, access environment variables, call external endpoints, or perform unrelated system actions.
Install Mechanism
No install spec or bundled code is present (instruction-only). Nothing is written to disk or downloaded during installation.
Credentials
The skill requires no environment variables, secrets, or credentials — proportional to a prompt-guidance utility.
Persistence & Privilege
always is false and the skill is user-invocable. It does not request persistent system presence or permissions to modify other skills or system settings.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install output-wrong-task - After installation, invoke the skill by name or use
/output-wrong-task - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of the output-wrong-task skill:
- Detects when output answers a related but incorrect task, often due to ambiguous prompts.
- Documents symptoms such as incomplete answers, solving adjacent problems, or misinterpreting compound requests.
- Provides actionable advice for crafting clearer prompts to ensure correct and focused responses.
- Lists steps users can take to anchor deliverable requirements and mitigate recurring "wrong task" outputs.
Metadata
Frequently Asked Questions
What is Output Wrong Task?
The model produces correct-looking output that addresses a different task than the one requested — typically a related but distinct interpretation of an ambi... It is an AI Agent Skill for Claude Code / OpenClaw, with 48 downloads so far.
How do I install Output Wrong Task?
Run "/install output-wrong-task" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Output Wrong Task free?
Yes, Output Wrong Task is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Output Wrong Task support?
Output Wrong Task is cross-platform and runs anywhere OpenClaw / Claude Code is available (macos, linux, windows).
Who created Output Wrong Task?
It is built and maintained by mvogt99 (@mvogt99); the current version is v1.0.0.
More Skills