The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs dependencies and fetches remote assets from the vendor's infrastructure.\n
Instructs installation of posthog-js and @posthog/types via standard package managers.\n
References an HTML snippet that downloads array.js from assets.i.posthog.com.\n- [REMOTE_CODE_EXECUTION]: The recommended integration method involves loading a remote script into the browser context.\n
The HTML snippet in references/js.md uses a dynamic loader to fetch and execute JavaScript from PostHog's CDN at runtime.\n- [COMMAND_EXECUTION]: The workflow involves executing local project scripts to validate changes.\n
references/basic-integration-1.2-revise.md instructs running linter or prettier scripts found in package.json on the files created or edited by the agent.\n- [DATA_EXFILTRATION]: The skill transmits user data and application events to external analytics endpoints.\n
Configures posthog.init with an api_host targeting us.i.posthog.com.\n
Captures event properties and user identity info (email, name, role) for transmission to the vendor's API.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface by ingesting project source files to generate a tracking plan.\n
Ingestion points: references/basic-integration-1.0-begin.md reads 10-15 files from the user project.\n
Boundary markers: The instructions do not define delimiters or ignore-embedded-instructions warnings for the project source code.\n
Capability inventory: The agent possesses capabilities to write local files (.posthog-events.json) and execute subprocesses (linting/building scripts) as specified in references/basic-integration-1.2-revise.md.\n
Sanitization: No explicit validation or escaping of the ingested project code is mentioned before the agent processes it to generate integration logic.

integration-javascript_web