The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from external .canvas files.
Ingestion points: The skill reads and parses existing .canvas files during the 'Add a Node' and 'Edit an Existing Canvas' workflows defined in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters mentioned to prevent the agent from following instructions potentially embedded within the JSON values (e.g., in a node's text field).
Capability inventory: The skill performs file system operations including reading (Read and parse the existing .canvas file) and writing (Write the updated JSON back to the file) as described in SKILL.md.
Sanitization: The skill lacks instructions for sanitizing or escaping the content of external files before the agent processes or interpolates them into its context.

json-canvas