The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes bash commands and curl to export project documentation and datasets from the Latitude V1 API, and it uses the latitude CLI to manage V2 projects and schema discovery.
[DATA_EXFILTRATION]: Accesses project environment variables, such as LATITUDE_V1_API_KEY and LATITUDE_PROJECT_ID, to authenticate requests against vendor-owned API endpoints (gateway.latitude.so and api.latitude.so). The skill explicitly instructs the agent never to print these secrets and to store them only in secure local secret stores or .env files.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it ingests existing application code and legacy prompt files to determine migration requirements.
Ingestion points: Reads local project files (requirements.txt, package.json, .env), V1 prompt files (.promptl), and JSON responses from the V1 API.
Boundary markers: Implements a mandatory "Plan, then wait" ground rule that requires the agent to present a detailed migration plan and receive human confirmation before executing any code changes.
Capability inventory: Includes the ability to perform network requests, read/write local files, and install new software packages via standard package managers (npm, pip).
Sanitization: Relies on a human-in-the-loop verification process where the user must audit the proposed migration plan before execution.
[DYNAMIC_EXECUTION]: Programmatically generates new Python and TypeScript modules by converting legacy templates into functional code that incorporates provider-specific SDK calls and telemetry instrumentation.

latitude-migrate