The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the download and installation of the latitude CLI binary. This is a standard procedure for the vendor's own tooling and is required for the onboarding workflow.
[COMMAND_EXECUTION]: Several shell commands are used to automate the configuration process, including the use of sed to manage .env files, jq for processing JSON metadata, and chmod to set permissions for the downloaded CLI binary.
[SAFE]: The skill demonstrates safe credential handling by using a temporary scratch file (~/.latitude-bootstrap.json) to move API keys into .env without echoing them to stdout or chat transcripts. It explicitly instructs the deletion of this file after use to prevent persistent local storage of secrets.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes external data from the Latitude API via latitude traces list.
Ingestion points: Structured JSON data from latitude traces list and latitude account bootstrap --schema is ingested into the agent's context.
Boundary markers: The skill uses structured JSON parsing (jq) to isolate specific fields, which acts as a technical boundary against the interpretation of the entire payload as instructions.
Capability inventory: The skill has the ability to write to local files (.env), delete/create projects via the CLI, and execute user-provided LLM code for verification.
Sanitization: Content is filtered through jq to select only specific data fields (like apiKey or projectSlug), which effectively sanitizes the input by ignoring non-requested JSON properties.

latitude-setup