Security audit
Security checks for vulnerabilities and agentic risk
This skill is a transparent utility for fetching coding-model rankings and pricing, with expected outbound requests but no credential use, persistence, or local system changes.
Before installing, be aware that running the script makes live requests to BenchLM and OpenRouter. It appears safe for normal use, but users who require strict offline operation or explicit network permission manifests may want that documented before use.
The skill clearly instructs users to run a script that performs outbound network requests, but the manifest does not declare any tool scope such as permissions or allowed-tools. This creates a transparency and policy-enforcement gap: an agent or user may invoke network-capable behavior without explicit declaration or review, increasing the risk of unintended external access.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- Cache benchmark data (rankings change frequently)
- Trust BenchLM pricing alone (use OpenRouter for live prices)
- Ignore tool-calling support for agentic use cases
- Recommend models without checking OpenRouter availability
- Mix up input vs output pricing (they differ significantly)
---
The skill describes live fetching from external APIs on every run and even states to always fetch fresh data, but it does not warn users that external requests will occur or clarify whether any user context could be transmitted. While the sample code only sends a User-Agent, the skill-level behavior still lacks disclosure and consent around network activity.
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
The comment describes the OpenRouter ID mapping as covering top coding models, suggesting a bounded set aligned with the skill's top-20 focus. The dictionary actually contains a much broader catalogue of models and aliases, which does not match the stated narrow scope of the comment.
The manifest says the skill supports IDE compatibility info for top coding models, which implies model-specific compatibility data. In this section, the script prints generalized OpenRouter setup instructions for many IDEs and states that all models with an OpenRouter ID are accessible in every major agentic coding IDE, but this is not computed from fetched model metadata and is only loosely tied to the ranked models.