The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes project-level configuration files including package.json, config.xml, capacitor.config.ts, and capawesome.config.json to determine framework versions and build settings. This ingestion of untrusted data from the user's repository creates a surface for indirect prompt injection.
Ingestion points: Configuration files are read and processed in references/live-updates.md, references/build-configuration.md, and references/live-updates-capacitor.md to guide the agent's deployment and build logic.
Boundary markers: The instructions do not employ explicit delimiters or system-level instructions to ignore potential commands embedded within the content of these files.
Capability inventory: The skill utilizes npx @capawesome/cli for network operations and cloud builds, npx trapeze for modifying local native projects, and openssl for handling sensitive certificate data.
Sanitization: There is no evidence of validation or sanitization of the values extracted from the configuration files before they are utilized in subsequent command executions or logic branching.
[CREDENTIALS_UNSAFE]: The skill provides instructions for managing sensitive authentication materials, such as Apple .p8 API keys, .p12 certificates, Android .jks keystores, and Google service account JSON files. While essential for the skill's primary function, these credentials are often passed as command-line arguments, which could lead to exposure in local shell histories. The skill correctly identifies the use of environment secrets for CI/CD workflows to mitigate this.
[EXTERNAL_DOWNLOADS]: The skill recommends adding a remote MCP server from mcp.capawesome.io and encourages the installation of external tools like the asc CLI and @trapezedev/configure. These resources are vendor-owned or recognized tools within the Capacitor and Cordova development communities.

capawesome-cloud