The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md include commands to start a background bridge server process using node ${CLAUDE_SKILL_DIR}/scripts/bridge-server.mjs &. This is a core component required to bridge the AI agent with the EasyEDA browser client.
[EXTERNAL_DOWNLOADS]: The skill provides URLs to download a required browser extension (run-api-gateway.eext) from vendor-controlled domains jlcext.com and jlc-ext.com. These are legitimate resources for the EasyEDA ecosystem.
[INDIRECT_PROMPT_INJECTION]: The skill has a large attack surface for indirect prompt injection because it is designed to read and process potentially complex electronic design files which may originate from untrusted third parties.
Ingestion points: Data enters the agent context through numerous API calls documented in SKILL.md and references/, such as eda.dmt_Project.getCurrentProjectInfo(), eda.sch_PrimitiveText.get(), and eda.sys_FileManager.getDocumentSource().
Boundary markers: Absent. The instructions do not currently require the agent to wrap design data in delimiters or use specific warnings when processing text from schematics or PCB layouts.
Capability inventory: The skill possesses significant capabilities including background process execution, local network communication, and filesystem access within the EDA client sandbox.
Sanitization: Absent. There are no instructions for sanitizing or escaping content retrieved from the design files before it is processed by the AI.

easyeda-api