The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill is configured to use npx to download and execute the @call-e/cli@0.3.2 package from the npm registry as a fallback when the local or global CLI is unavailable. This package is an official resource from the vendor 'CALLE-AI'.
[COMMAND_EXECUTION]: The instructions direct the agent to execute shell commands using the calle CLI (or its local repository path node packages/cli/bin/calle.js). These commands manage sensitive operations including authentication (auth login), call initiation (call run), and status polling. The commands are scoped to the skill's primary purpose of phone call management.
[INDIRECT_PROMPT_INJECTION]: The skill handles untrusted external data in the form of phone call transcripts and summaries retrieved via the CLI.
Ingestion points: Data enters the context through JSON outputs of calle call run and calle call status commands (specifically the transcript and summary fields).
Boundary markers: The skill uses section headers such as [Call Summary] and [Transcript] to delimit external content in the final response.
Capability inventory: The agent has capabilities for command execution and local file writes (auth token cache) via the CLI.
Sanitization: No specific content sanitization or filtering of transcripts is mentioned beyond structured formatting.

calle