← Back to Skills Marketplace
79
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install agency-evidence-collector
Description
Expert AI agent specializing in evidence collector. From The Agency (github.com/msitarzewski/agency-agents).
README (SKILL.md)
evidence collector
Identity & Style
🧠 Your Identity & Memory
- Role: Quality assurance specialist focused on visual evidence and reality checking
- Personality: Skeptical, detail-oriented, evidence-obsessed, fantasy-allergic
- Memory: You remember previous test failures and patterns of broken implementations
- Experience: You've seen too many agents claim "zero issues found" when things are clearly broken
Core Mission
🔍 Your Core Beliefs
"Screenshots Don't Lie"
- Visual evidence is the only truth that matters
- If you can't see it working in a screenshot, it doesn't work
- Claims without evidence are fantasy
- Your job is to catch what others miss
"Default to Finding Issues"
- First implementations ALWAYS have 3-5+ issues minimum
- "Zero issues found" is a red flag - look harder
- Perfect scores (A+, 98/100) are fantasy on first attempts
- Be honest about quality levels: Basic/Good/Excellent
"Prove Everything"
- Every claim needs screenshot evidence
- Compare what's built vs. what was specified
- Don't add luxury requirements that weren't in the original spec
- Document exactly what you see, not what you think should be there
How to Activate
Reference this agent by name or specialty when you need its expertise.
Usage Guidance
This skill appears to be a legitimate QA/evidence-collector, but it has two practical issues you should verify before enabling it: (1) inspect the referenced script ./qa-playwright-capture.sh (and any Playwright test files) to ensure they don't exfiltrate data, spawn unexpected processes, or call external endpoints; (2) confirm whether your agent runtime actually has Playwright/node and the necessary scripts available — the skill's metadata doesn't declare these dependencies. Also be aware that when invoked this skill will run shell commands and read local project files (public/qa-screenshots, resources/views, test-results.json), so only enable it in environments where you trust those actions. If you want to proceed, request the author add explicit dependency and safety metadata (required binaries, a manifest of scripts to run, and a note about network behavior) or provide the capture script contents for review.
Capability Analysis
Type: OpenClaw Skill
Name: agency-evidence-collector
Version: 1.0.0
The skill bundle defines a QA agent persona that is instructed to execute shell commands and local scripts for evidence collection. Specifically, AGENTS.md contains a 'Mandatory Process' requiring the execution of a local script `./qa-playwright-capture.sh` against a local endpoint (http://localhost:8000) and performing file system operations like `ls` and `grep` to verify implementations. While these actions are aligned with the stated QA purpose, the use of shell execution and local network access constitutes a risky capability profile that warrants a suspicious classification under the provided criteria.
Capability Assessment
Purpose & Capability
The name/description (evidence collector / QA) match the included SKILL.md/AGENTS.md content. However, AGENTS.md's mandatory Step 1 requires running a local Playwright capture script (./qa-playwright-capture.sh) and inspecting local files (public/qa-screenshots, resources/views, test-results.json). The skill declares no required binaries, env vars, or install steps — a mismatch between stated metadata and the actions the instructions expect.
Instruction Scope
Instructions direct the agent to run shell commands (Playwright capture script, ls, grep, cat) and to read specific local paths. These actions are coherent with QA work but grant the agent permission to execute arbitrary local scripts and read project files. The SKILL.md does not show the contents of the referenced ./qa-playwright-capture.sh script or otherwise constrain what will be executed or where data may be sent.
Install Mechanism
There is no install spec and no code files that would be downloaded/executed by an installer; this lowers supply-chain risk. However, because the instructions assume tooling (Playwright, a capture script), the runtime environment must already have those tools — which the manifest does not declare.
Credentials
The skill requests no environment variables or credentials, which is reasonable for a local QA agent. Still, the instructions imply access to local filesystem paths and to run a script that likely requires Node/Playwright; those runtime requirements are not declared, so the lack of declared secrets is not sufficient reassurance.
Persistence & Privilege
The skill is not marked always:true, and it doesn't request to modify other skills or system-wide agent settings. It appears limited to being invoked by name and performing local QA actions when invoked.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agency-evidence-collector - After installation, invoke the skill by name or use
/agency-evidence-collector - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of agency-evidence-collector.
- Introduces an expert agent focused on visual evidence collection and reality checking.
- Emphasizes screenshot-based verification and skepticism of unproven claims.
- Provides clear beliefs: prioritize visual proof, expect to find multiple issues, and avoid adding extra requirements.
- Designed for quality assurance roles, documenting test failures and discrepancies in implementations.
Metadata
Frequently Asked Questions
What is test?
Expert AI agent specializing in evidence collector. From The Agency (github.com/msitarzewski/agency-agents). It is an AI Agent Skill for Claude Code / OpenClaw, with 79 downloads so far.
How do I install test?
Run "/install agency-evidence-collector" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is test free?
Yes, test is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does test support?
test is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created test?
It is built and maintained by zhouqkt (@zhouqkt); the current version is v1.0.0.
More Skills