100Critical
The skill permits arbitrary command execution and URL navigation,
Behavior Analysis
Description verified — behavior matches claim
Claims to do
Slack Automation: Interact with Slack workspaces to check messages, extract data, and automate common tasks.
Actually does
This skill uses the `agent-browser` tool to perform browser automation on `https://app.slack.com`. It can connect to an existing browser session (port 9222) or open a new one. It interacts with the Slack web interface by taking DOM snapshots, clicking elements, filling forms, pressing keys, waiting, taking screenshots, scrolling, and extracting text, URLs, and titles.
Skill Info
Namevercel-labs/agent-browser/slack
Registrygithub
Versiondb29d5feade8
PURLpkg:github/vercel-labs/agent-browser@db29d5feade8?skill=slack
Stars29,298
Installs14,100
Source
SHA-25616817a09c54f...
SHA-128c83637fdd5...
MD57e2dcba73cc5...
TLSH9df150d5200e...
ssdeep192:X0FvJGbd...
Size8,032 bytes
Install
Assessments (1)
AI Agent Traps ↗Broad Browser Automation & Command Executioncriticalcommand execution
The skill permits execution of arbitrary `agent-browser` commands via Bash, granting full control over a browser instance, including navigation to any URL and interaction with page content.
100% confidenceline 4
allowed-tools: Bash(agent-browser:*), Bash(npx agent-browser:*)
Sensitive Data Collection (Exfiltration Risk)highdata exfiltration
The skill enables collection of sensitive data from any webpage via `get text`, `snapshot --json`, and `screenshot` commands, posing a risk for data exfiltration.
100% confidenceline 195
agent-browser snapshot --json > output.json
Credential Harvesting Capabilityhighcredential harvesting
The agent can navigate to login pages and extract credentials using `get text` from input fields or `screenshot` commands, posing a risk for credential harvesting.
90% confidenceline 188
agent-browser get text @e_message_ref
Reconnaissance via Arbitrary Web Navigationmediumreconnaissance
The `agent-browser open <url>` command allows navigation to arbitrary URLs, enabling reconnaissance of internal or external web resources beyond the skill's stated purpose.
100% confidenceline 20
agent-browser open https://app.slack.com
Dependencies (1)
agent-browser.claude-plugin/marketplace.jsonnot scanned
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/github/vercel-labs/agent-browser/slack)HTML
<a href="https://mondoo.com/ai-agent-security/skills/github/vercel-labs/agent-browser/slack"><img src="https://mondoo.com/ai-agent-security/api/badge/github/vercel-labs/agent-browser/slack.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/github/vercel-labs/agent-browser/slack.svgSecure your AI agents
Skills can read files, run commands, and access credentials. Mondoo helps organizations manage the security risks of AI agent skills across their entire fleet.
- Continuous scanning of skills across all registries
- Policy enforcement before skills reach your agents
- Integration with your existing security workflow