← Back to Skills Marketplace
xisen-w

Examine Sandbox

by Awassi · GitHub ↗ · v2.0.0 · MIT-0
cross-platform ⚠ suspicious
92
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install examine-sandbox
Description
Use this skill when the user wants to check what data their shared agent can access, inspect what's being shared, review privacy, or see what guests will see...
README (SKILL.md)

Examine Sandbox

Inspect exactly what data and capabilities are included in shared links.

Prerequisites

  • PULSE_API_KEY must be set
  • Base URL: https://www.aicoo.io/api/v1

Core Workflow

Step 1: List network state

curl -s -H "Authorization: Bearer $PULSE_API_KEY" \
  "https://www.aicoo.io/api/v1/os/network" | jq .

Review:

  • shareLinks
  • visitors
  • contacts

Step 2: Check context size/scope

curl -s -H "Authorization: Bearer $PULSE_API_KEY" \
  "https://www.aicoo.io/api/v1/os/status" | jq .

Step 3: Search for sensitive content

# financial
curl -s -X POST "https://www.aicoo.io/api/v1/os/notes/search" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"revenue pricing confidential"}' | jq .

# credentials/personal info
curl -s -X POST "https://www.aicoo.io/api/v1/os/notes/search" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"password API key credentials"}' | jq .

Step 4: Report findings

Summarize:

  1. how many active links and their scopes
  2. notes/calendar permission levels
  3. visitor activity
  4. sensitive hits inside shared scope
  5. risk actions (downgrade/revoke)

Step 5: Restrict access if needed

# narrow scope
curl -s -X PATCH "https://www.aicoo.io/api/v1/os/share/{linkId}" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"scope":"folders","folderIds":[5,12]}' | jq .

# downgrade notes access
curl -s -X PATCH "https://www.aicoo.io/api/v1/os/share/{linkId}" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"notesAccess":"read"}' | jq .

# revoke
curl -s -X DELETE "https://www.aicoo.io/api/v1/os/share/{linkId}" \
  -H "Authorization: Bearer $PULSE_API_KEY" | jq .

Search Categories

Category Terms Risk
Financial revenue, burn rate, pricing, salary medium
Credentials password, token, key, secret critical
Personal phone, address, SSN, private high
Legal contract, NDA, agreement high
Usage Guidance
This skill appears to do what it says (audit and manage share links on aicoo.io), but note two issues: (1) the runtime docs require PULSE_API_KEY even though the registry metadata lists no required env vars — don't assume the registry metadata is complete; (2) the API key has high privilege (can list and revoke links and change access). Only provide a PULSE_API_KEY that is scoped/minimized for auditing (use a temporary or read-only token if available), verify the aicoo.io endpoint and that you trust the skill author (no homepage, unknown owner), and rotate/revoke the key after use. If you need higher assurance, request the author update the package metadata to explicitly declare PULSE_API_KEY and provide an official homepage or documentation.
Capability Analysis
Type: OpenClaw Skill Name: examine-sandbox Version: 2.0.0 The examine-sandbox skill is designed to audit shared links and data exposure on the aicoo.io platform. It is classified as suspicious because it includes explicit instructions for the agent to programmatically search for and retrieve highly sensitive information, such as 'password', 'API key', and 'revenue', across user notes (SKILL.md, reference/API.md). While these capabilities are plausibly needed for the stated purpose of a security audit, the automated discovery of credentials represents a high-risk behavior that could be exploited, and the skill lacks sufficient safeguards against the agent exposing these secrets in its output.
Capability Tags
requires-sensitive-credentials
Capability Assessment
Purpose & Capability
The SKILL.md clearly targets auditing and managing shared links on aicoo.io and the curl commands reflect that purpose (list links, search notes, patch/revoke links). Requiring an API key for that service is reasonable for the stated capability. However, the registry metadata claims no required env vars or primary credential while the runtime docs explicitly require PULSE_API_KEY — an inconsistency between declared requirements and actual runtime needs.
Instruction Scope
Instructions are specific and constrained to the aicoo.io API endpoints (GET /os/network, POST /os/notes/search, PATCH/DELETE /os/share/{linkId}). They do not instruct reading arbitrary local files or other environment variables. They do include write operations (PATCH, DELETE) which can modify or revoke share links; this is coherent with 'restrict access' actions but is higher privilege and should be used carefully.
Install Mechanism
This is an instruction-only skill with no install spec or code files, so it does not write code to disk or fetch remote packages. That's low installation risk.
Credentials
The SKILL.md requires a single env var PULSE_API_KEY (used for Authorization: Bearer) which is proportionate to the API operations described. The concern is that the registry metadata lists no required env vars or primary credential — a metadata omission that could mislead users. Also, the PULSE_API_KEY likely grants listing/modifying/revoking share links (sensitive capabilities); users should ensure the key is least-privilege and scoped appropriately.
Persistence & Privilege
The skill is not always-enabled, has no install-time persistence, and does not request system-wide configuration changes. It runs ephemeral curl commands at runtime. Autonomous invocation is allowed (platform default) but does not by itself increase risk here.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install examine-sandbox
  3. After installation, invoke the skill by name or use /examine-sandbox
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v2.0.0
Initial publish from pulse-skills
Metadata
Slug examine-sandbox
Version 2.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Examine Sandbox?

Use this skill when the user wants to check what data their shared agent can access, inspect what's being shared, review privacy, or see what guests will see... It is an AI Agent Skill for Claude Code / OpenClaw, with 92 downloads so far.

How do I install Examine Sandbox?

Run "/install examine-sandbox" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Examine Sandbox free?

Yes, Examine Sandbox is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Examine Sandbox support?

Examine Sandbox is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Examine Sandbox?

It is built and maintained by Awassi (@xisen-w); the current version is v2.0.0.

💬 Comments