T09 · Insecure Skill Coding Practices
- Location
SKILL.md:188- Finding
Caller-Controlled URL Fetching Enables Server-Side Request Forgery
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
The skill is a coherent service-hub helper, but its examples expose local files, arbitrary outbound requests, and service keys without enough scoping or safety guidance.
Review before installing. Use this only in a controlled environment, pin and verify dependencies, avoid importing client code from mutable workspaces, require wss:// for remote hubs, never log keys, and do not publish file or URL-fetch services without allowlists, path containment, authentication, and explicit user approval.
SKILL.md:188Caller-Controlled URL Fetching Enables Server-Side Request Forgery
SKILL.md:160Path Traversal in Published File Service Can Expose Local Data
SKILL.md:478Plaintext WebSocket Transport Can Expose Service Data and Authorization Keys
SKILL.md:14Unpinned and Unreviewed Runtime Dependencies Create Supply-Chain Exposure
The trigger patterns are very broad, matching common phrases like use.*service, query.*data, and fetch.*data, which are likely to activate this skill for many unrelated user requests. In an agent ecosystem, overbroad activation can unnecessarily grant service-discovery, network, and data-sharing behaviors in contexts where the user did not explicitly ask to use the hub.
The documentation instructs users to prepend WORKSPACE_DIR to sys.path, causing Python to import modules from an arbitrary workspace location. In an agent workspace, that can result in loading attacker-planted code through import side effects, effectively turning documentation into a pattern for arbitrary code execution beyond the marketplace skill’s core role.
The examples show exposing local files as services and fetching arbitrary remote URLs, but they do not clearly warn that these actions can disclose local sensitive data or transmit data to external systems. Because this is documentation for agent behavior, omission of privacy and egress warnings materially increases the chance of unsafe deployment and accidental leakage.
The skill includes a generic fetch_data(url=...) example that allows arbitrary outbound HTTP requests unrelated to a narrowly scoped service marketplace helper. In an agent setting, this expands capability into general network access, enabling SSRF-like access to internal services, data exfiltration, or use as an unintended proxy if consumers pass attacker-controlled URLs.
The key-management example prints part of the authorization key and advises consumers to save keys without giving handling requirements. Even partial key disclosure and casual storage guidance normalize insecure secret practices, increasing the risk of credential leakage through logs, transcripts, screenshots, or insecure local persistence.
The provider template instructs users to use an English service name and English description, which is a language constraint stated in natural language. The document does not present this as optional, offer alternatives, or justify the restriction as region- or protocol-specific.
Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,