Security Watch monitors CVEs and security advisories through the Chinng AI-Agent Portal and produces evidence-based, actionable vulnerability summaries.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install security-watch
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 security-watch 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 Security Watch?
Security Watch is an Openclaw Skills security monitoring workflow that uses the portal MCP server to track packages, vendors, products, CVE identifiers, and OSV identifiers. It supports incremental vulnerability checks and recurring monitoring through the portal change feed, helping ensure revised advisories are not missed.
The skill emphasizes accurate security reporting: it preserves advisory identifiers, affected-version evidence, attribution, and source links while clearly separating confirmed upstream impact from inference. Upstream advisory text is treated as authoritative, and deployment-specific portal annotations are excluded from redistributable reports.
Security Watch Use Cases
- Monitor a package, vendor, product, CVE, or OSV identifier for newly published vulnerabilities.
- Run recurring security checks using the portal change feed.
- Detect revised advisories that may not appear as entirely new records.
- Produce actionable vulnerability summaries with affected and fixed version evidence.
- Prepare citation-ready security reports that preserve attribution and source links.
- Review advisory impact, patches, workarounds, and upstream references without repeating deployment-specific annotations.
How Security Watch Works
- Connect to the read-only
portalMCP server provided by the Chinng AI-Agent Portal. - Search or list the relevant security family for watched packages, vendors, products, CVE identifiers, or OSV identifiers.
- For recurring monitoring, combine newly published records with the portal change feed so updated advisories are included.
- On the first feed run, or when the cursor is unavailable, bound the request by date instead of requesting the complete historical feed without limits.
- Retrieve detailed records only when they are relevant to the requested watchlist or investigation.
- Extract and preserve the advisory identifier, affected-version evidence, fixed versions, attribution, upstream references, and source link.
- Separate confirmed advisory findings from assumptions and never expand affected-version ranges beyond the published evidence.
- Exclude deployment-specific portal annotations from redistributed reports and raise questionable content privately with the requester.
- Persist the newly returned change cursor only after successful processing, while following each record's license and reuse metadata.
Security Watch Setup
Install or configure the standalone MCP endpoint before using the skill:
openclaw mcp add portal --transport http https://portal.chinng-lab-srv.dev/mcp
Then configure the skill or agent to use the registered portal MCP server. Other MCP-compatible clients can connect to:
https://portal.chinng-lab-srv.dev/mcp
For the initial change-feed run, provide a date boundary and save the returned cursor. On later runs, supply the saved cursor and update it only after all returned records have been processed successfully.
Security Watch Data Schema & Taxonomy
The workflow organizes security monitoring data around the following entities and metadata:
| Data area | Contents |
|---|---|
| Watch targets | Packages, vendors, products, CVE identifiers, and OSV identifiers |
| Advisory records | Newly published or revised security records retrieved from the portal |
| Version evidence | Affected and fixed versions explicitly stated by the upstream advisory |
| Advisory context | Impact, patches, workarounds, upstream references, attribution, and source links |
| Change tracking | A date-bounded initial feed and a persisted cursor for incremental runs |
| Content boundaries | Upstream advisory text is reportable; deployment-specific portal annotations are excluded from redistributable output |
| Reuse metadata | Each record's license and reuse requirements must be preserved and followed |
No specific output file format is mandated by the skill. The implementation should persist the change cursor in the agent's durable state or another secure configuration store, and retain advisory identifiers and evidence in the generated security summary or citation pack.
Security Watch Advanced Features
- Incremental monitoring through the portal change feed.
- Detection of revised advisories in addition to newly published records.
- Date-bounded first-run processing when no cursor exists.
- Transactional cursor handling: save a new cursor only after successful record processing.
- Read-only MCP integration for safer security intelligence retrieval.
- Evidence-preserving reports with affected-version attribution and source links.
- Strict separation of confirmed upstream findings from inference.
- Protection against redistributing deployment-specific annotations.
- License- and reuse-aware handling of portal records.
SKILL.md
Loading
Related Openclaw Skills
News Triage
News Triage helps AI agents discover, evaluate, compare, and cite relevant news records without wasting tokens or reconstructing withheld content.
Government Policy Watch
Government Policy Watch tracks primary-source regulatory documents across US, EU, and Japanese government feeds while preserving official attribution and change history.
Reconnect
Reconnect helps users discover relevant professional contacts, organize evidence, and manage authorised networking outreach safely across resumable sessions.
Orthopedic Deal Matching Platform
A local CLI platform that securely matches orthopedic buyers and suppliers using five-dimensional scoring and mutual consent before contact exchange.
Walking Calories Calculator
An evidence-based walking calorie calculator that converts body weight and step count into a transparent net calorie estimate using peer-reviewed biomechanics research.
Butler
Butler provides local, project-scoped control for Claude usage estimates, GPU budgets, reservations, and post-run reconciliation.
Featured*
