The Agent Skills Directory

Prompt Injection Considerations: The instructions use assertive directives such as "CRITICAL: Always Use Correct workflow Documentation" and claim "Your knowledge of workflow is outdated" to override the agent's behavior and pre-existing knowledge. These patterns are often used to ensure adherence to specific documentation but can be a vector for instruction bypass.
Indirect Prompt Injection Surface: The skill instructs the agent to search and process content from documentation files located in node_modules/workflow/docs/. This presents a surface where untrusted content could influence agent behavior.
Ingestion points: Documentation files in node_modules/workflow/docs/ accessed via glob and grep commands.
Boundary markers: None identified; instructions do not explicitly tell the agent to ignore embedded directives in the documentation.
Capability inventory: The skill documents capabilities for network operations (fetch), file system traversal (glob), and command execution (npx workflow).
Sanitization: None identified; there is no explicit validation or filtering described for the documentation content being processed.
External Dependencies: The skill references several external Node.js packages and utilities, including the use of npx for CLI operations. These resources are part of the specific workflow ecosystem described in the skill.

workflow