The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill documents the use of node:child_process.fork() to execute worker scripts for formula and Base processing. While this is a standard architectural pattern for the library to offload heavy computations, it involves dynamic execution of local scripts.
Evidence: In SKILL.md and references/node-pro-integration.md, code snippets demonstrate forking formula-worker.js and base-worker.js resolved via import.meta.url.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of workbook data from external JSON snapshots, creating an attack surface for malicious instructions embedded in document data (e.g., within formulas).
Ingestion points: Documented readFile operations in references/node-common-tasks.md load workbook data from the local file system into the agent's context.
Boundary markers: The provided examples do not include explicit boundary markers or instructions to ignore embedded content when processing external data.
Capability inventory: The skill utilizes the UniverFormulaEnginePlugin for calculation and node:child_process.fork() for worker execution, and it has file system write capabilities via writeFile.
Sanitization: No explicit sanitization or strict schema validation is showcased in the examples for the ingested JSON data.

univer-node-backend