The Agent Skills Directory

[Data Exposure & Exfiltration] (SAFE): The scripts scripts/graphql-client.sh and scripts/schema-introspection.py perform network requests to the GitHub GraphQL API. This is consistent with the skill's stated purpose of providing GraphQL tooling. No access to sensitive local files (like SSH keys or AWS credentials) was found.
[Unverifiable Dependencies & Remote Code Execution] (SAFE): The skill references standard, industry-recognized libraries such as requests in Python and the Apollo suite in Node.js. There are no patterns of piping remote content to a shell or downloading executable code from untrusted sources.
[Indirect Prompt Injection] (LOW): The script scripts/graphql-client.sh interpolates a username variable directly into a GraphQL query string. This creates a surface for GraphQL injection if the input is malicious, though in the context of a development tool, this is a common functional pattern.
Ingestion points: username argument in scripts/graphql-client.sh.
Boundary markers: None present in the script's query construction.
Capability inventory: Network access via curl.
Sanitization: No sanitization of the username variable before interpolation.
[Command Execution] (LOW): The skill includes a bash script that uses curl and jq. These are standard utilities used for their intended purpose of API communication and JSON parsing.

graphql