A specialized diagnostic tool to verify the completeness and integrity of workspace snapshots before they are exported or consumed by AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install read-only-workspace-audit
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install read-only-workspace-audit using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Read-Only Workspace Audit?
The Read-Only Workspace Audit is a technical utility designed to evaluate the depth and accuracy of imported Campus Copilot workspace snapshots. Within the ecosystem of Openclaw Skills, this tool ensures that academic data—sourced from platforms like Canvas, Gradescope, and EdStem—is sufficiently detailed to support cited AI responses or high-fidelity exports. It operates on a strict read-only principle, preserving the integrity of the original evidence while providing a transparent audit trail for local-first academic workflows.
By leveraging this skill, developers and researchers can determine exactly what evidence is available within a snapshot without risking the corruption of session truth. It bridges the gap between raw data collection and Model Context Protocol (MCP) consumption, making it a cornerstone for those building reliable, local-first AI applications using Openclaw Skills.
Read-Only Workspace Audit Use Cases
- Verifying if a workspace snapshot contains enough evidence for export to cited AI platforms.
- Auditing local academic data from Canvas, Gradescope, EdStem, or MyUW without modifying the source.
- Establishing a strict audit trail for local consumers within the Openclaw Skills framework.
- Testing the readiness of MCP servers for Codex or Claude Code integrations.
How Read-Only Workspace Audit Works
- Identify the path to an imported Campus Copilot workspace snapshot or exported artifact.
- Run a summary command to visualize the current state of the workspace and open sites.
- Perform site-specific audits to verify that evidence from platforms like Canvas or MyUW is present.
- Initialize a read-only MCP server to expose the snapshot data to external AI agents.
- Evaluate the richness of the snapshot to determine if it meets the requirements for cited AI consumption.
Read-Only Workspace Audit Setup
To begin auditing your snapshots within the Openclaw Skills environment, use the following commands:
# Generate a summary of the current snapshot
campus-copilot summary --snapshot <path>
# Audit a specific site within the snapshot
campus-copilot site --snapshot <path> --site <canvas|gradescope|edstem|myuw>
# Start the read-only MCP server for snapshot consumption
pnpm --filter @campus-copilot/mcp-server start
Read-Only Workspace Audit Data Schema & Taxonomy
The skill organizes audit data and metadata according to the following structure:
| Data Component | Description |
|---|---|
| Workspace Snapshot | The root JSON or directory structure containing the captured academic state. |
| Site Evidence | Categorized data points mapped to specific academic portals (e.g., Gradescope). |
| MCP Artifacts | JSON-formatted files optimized for Model Context Protocol discovery. |
| Audit Logs | Local-first logs detailing the richness and completeness of the snapshot. |
Read-Only Workspace Audit Advanced Features
- Site-scoped MCP servers for granular, snapshot-only read access.
- Multi-agent support for Codex and Claude Code via pre-configured JSON examples.
- Repo-local verification to ensure snapshots are ready for cited AI processing.
- Local-first architecture that prioritizes academic privacy and data ownership within Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Site MCP Consumer
A bridge for connecting site-specific, read-only academic data snapshots to local AI agent runtimes via the Model Context Protocol.
Cognitive Topology
A sophisticated task-splitting engine that transforms linear agent reasoning into a parallel, multi-branched cognitive architecture.
OpenDataLoader PDF
An advanced AI-ready PDF parser designed to convert complex documents into LLM-friendly formats like Markdown and JSON for RAG pipelines.
Auto Purchase Executor Claw
A rule-driven procurement engine that automates the entire purchase cycle from stock monitoring to payment execution.
Easy Subtitles
A powerful AI video editing skill that automates caption generation and cloud rendering for content creators.
Benchmark Filter
A high-precision filtering tool designed to evaluate the actual business value of creator benchmarks and one-person business models.
Featured*
