T09 · Insecure Skill Coding Practices
- Location
SKILL.md:14- Finding
Workspace-Controlled Issuer URL Permits Arbitrary Outbound Requests
- Content
View full analysis
Vulnerability Details
File Location:
SKILL.md, line 14
Vulnerability Type: Server-Side Request Forgery through an unvalidated issuer URL
Risk Level: MediumVulnerable Code Snippet:
markdown Alias for `/whoareyou`. Follow that skill's instructions exactly — read your DID and `wayidIssuer` from `{openclaw-path}/workspace/wayid.json` (or `workspace-<agentId>/wayid.json` for named agents), `GET {wayidIssuer}/api/v1/agent/{bare-id}/card` (the bare 24-char identifier after `wayid:agent:`; default base URL `https://way.je` if `wayidIssuer` is missing), and render the card with the template defined in `/whoareyou`.Technical Analysis
The Skill directs the agent to read
wayidIssuerfrom a workspace JSON file and use it directly as the origin of an outbound HTTP request. It does not require HTTPS, restrict the issuer to an allowlist, validate the URL scheme and host, or reject loopback, link-local, and private network destinations.If an attacker can modify the applicable
wayid.json, they can substitute an arbitrary URL. Invoking/whowould then cause the agent to send a request to the selected destination. This creates a limited server-side request forgery primitive and discloses the agent's bare identifier in the request path.Attack Path
- An attacker obtains the ability to create or modify the workspace's
wayid.json. - The attacker sets
wayidIssuerto an attacker-controlled origin or an internal network endpoint. - A user invokes
/who, or otherwise triggers this identity Skill. - The agent constructs
GET {wayidIssuer}/api/v1/agent/{bare-id}/card. - The selected server receives the request and the agent identifier. If internal destinations are reachable, the request can also be used for network probing or interaction with an internal HTTP service.
Impact Assessment
Exploitation can disclose the agent's bare WayID identifier to an unintended party and make HTTP ...[truncated 469 chars]
- An attacker obtains the ability to create or modify the workspace's
- Remediation
View remediation
Remediation Suggestions
- Require
wayidIssuerto use HTTPS. - Parse the URL with a strict URL parser rather than concatenating an unchecked string.
- Maintain an explicit allowlist of trusted issuer origins.
- Reject URLs containing credentials, unexpected ports, redirects to untrusted origins, or non-HTTP schemes.
- Resolve the destination and reject loopback, link-local, private, multicast, and otherwise reserved IP ranges for both IPv4 and IPv6.
- Revalidate every redirect target and protect against DNS rebinding by validating the resolved connection destination.
- Ask for explicit user approval before contacting an issuer other than the documented default.
- Avoid transmitting the identifier until the destination has passed all validation checks.
- Require
