The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill facilitates the creation and storage of executable JavaScript and JSX code blocks within the Petri net document. These snippets define transition logic, stochastic rates, and custom SVG visualizations.\n
Evidence: In skill-api.ts, the addPlace, addTransition, and addDifferentialEquation methods accept and store raw strings for visualizerCode, lambdaCode, and code fields.\n
Evidence: SKILL.md provides examples of functional code meant to be embedded, such as export default Lambda((tokens, params) => params.beta);.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes Petri net definitions from external Automerge URLs, which serves as an ingestion point for untrusted data that includes executable logic.\n
Ingestion points: getPetriNet(url) in skill-api.ts loads documents from user-provided URLs.\n
Boundary markers: None. The skill does not differentiate between user-provided logic and logic loaded from an external document.\n
Capability inventory: The agent can read, modify, and create Petri nets with logic fields. The logic is rendered in the tool's React component (tool.tsx) via the @hashintel/petrinaut library.\n
Sanitization: The skill does not validate or sanitize the content of the code strings before they are persisted or processed.

petrinaut