Security audit
Security checks for vulnerabilities and agentic risk
The skill is a project-management API helper, but it requests broad local tool access and can change or send project data without enough scoping or user-control guidance.
Review this skill before installing in a workspace with real project data. Use a least-privilege API key, avoid untrusted callback URLs, and require the agent to ask before archiving, deleting, bulk-updating, or changing workflow state. The artifact does not show malware or hidden execution, but its local exec/write authority and weak mutation safeguards are broader than the API purpose needs.
The invocation description is vague and overbroad, describing flexible automation and multiple scenarios without clear boundaries on what the skill should or should not do. Ambiguous scope is dangerous for agentic systems because it increases the chance the model interprets user requests expansively and takes actions beyond intended project-management API operations.
The skill advertises a GraphQL project-management integration but also requests exec, which materially expands its authority beyond what is needed for the stated purpose. This increases the blast radius of prompt misuse or downstream instruction injection, because a workflow that should only call an external API could instead run local shell commands on the agent host.
The callback URL parameter enables sending results to an external endpoint, but the skill provides no privacy or trust warning about exfiltrating project data to third parties. In practice, a malicious or mistaken callback destination could receive sensitive issue contents, team metadata, comments, or identifiers without adequate user awareness.
The skill documents destructive actions such as updating state and archiving items without requiring confirmation or warning about irreversible or disruptive effects. In an agent context, that can lead to accidental modification of live project data, workflow disruption, or loss of visibility when a mistaken request is executed automatically.
The documentation mixes a project-management API skill with unrelated file handling and command-execution troubleshooting, which can normalize broader behavior than users expect. That mismatch makes it easier for an agent or operator to justify unsafe local actions under the cover of an API integration, increasing the chance of unintended filesystem access or command execution.
No suspicious patterns detected.
