other
Warning
- Location
- SKILL.md:80
- Finding
- Unverified External Agent Registration and Recurring Telemetry## Vulnerability Details **File Location**: `SKILL.md:80-86` **Vulnerability Type**: External Agent Registration and Data Disclosure **Risk Level**: Medium **Evidence**: ```bash curl -s -X POST https://boat-atlas-spa-flexible.trycloudflare.com/governance/register \ -H "Content-Type: application/json" \ -d '{"agent_name":"你的Agent名称"}' ``` The service automatically returns a dedicated Pro key and rejects duplicate names. Registration grants an initial base score of 15 points. ### Technical Analysis The Skill directs users or Agents to submit an Agent identifier to a remotely controlled `trycloudflare.com` tunnel. It also describes a 12-hour heartbeat model and additional health, trust-score, audit, and dashboard interactions with the same host. A Cloudflare Tunnel hostname can route requests to a backend that is not included in the audited project and can be modified independently after the Skill has been reviewed. The repository contains no local implementation of the advertised trust engine, registration service, VPAV validation, or trust calculations. Consequently, the server's data collection, credential issuance, response integrity, and future behavior cannot be verified through static analysis of this package. The Skill does not provide verifiable service ownership, an API specification, a privacy policy, data-retention rules, telemetry details, or credential lifecycle documentation. Although the health-check key is presented as public rather than confidential, it does not authenticate the remote operator or provide meaningful caller isolation. ### Attack Path 1. A user or Agent follows the registration command in `SKILL.md`. 2. The request sends an Agent-selected identifier to the external tunnel and necessarily exposes connection metadata such as the source IP address, request time, and HTTP client characteristics. 3. The external service returns a Pro key whose scope, expiration, storage requirements, and re ...[truncated 1219 chars]
- Remediation
- ## Remediation Suggestions 1. Replace the temporary tunnel hostname with a stable domain owned and controlled by the responsible organization. 2. Publish verifiable service ownership information and a documented API specification. 3. Clearly identify every field and metadata item collected by registration, audit, VPAV, health, and trust-score endpoints. 4. Provide privacy, retention, deletion, and third-party sharing policies before requesting registration. 5. Require explicit user consent before transmitting an Agent identifier or enabling recurring communication. 6. Make heartbeat and audit requests opt-in rather than presenting them as an automatic operational requirement. 7. Document returned credential scopes, expiration, secure storage, rotation, and revocation procedures. 8. Publish the server implementation or a reproducible verification mechanism for trust-score and VPAV results. 9. Clarify that this package is documentation for a remote service and does not itself contain a locally auditable trust-engine implementation.