The Agent Skills Directory

[SAFE]: The skill interacts exclusively with the official Goldsky ecosystem, including the goldsky CLI and the @goldskycom/cli NPM package. All external resources align with the vendor's identity (goldsky-io).
[COMMAND_EXECUTION]: The skill uses shell commands to manage subgraphs via the goldsky CLI. These commands include list, log, start, pause, deploy, and delete, which are standard for the tool's intended use.
[DATA_EXFILTRATION]: No evidence of unauthorized data transfer or exfiltration to non-whitelisted domains was found. Access is limited to project data and logs required for diagnostics.
[PROMPT_INJECTION]: The skill handles subgraph logs, which constitute a surface for indirect prompt injection. However, the instructions constrain the agent to diagnostic parsing and error identification, minimizing the risk of instructions in data being obeyed.

subgraph-doctor