The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill generates code designed to handle sensitive credentials including JPM_CLIENT_ID, JPM_PRIVATE_KEY_PATH, and JPM_CERT_THUMBPRINT. It also interacts with .env files to append placeholders for these values. These actions are transparently documented as part of the authentication setup process.
[COMMAND_EXECUTION]: The skill performs file system operations to read project manifest files (e.g., package.json, pom.xml, requirements.txt) to detect the development environment and writes generated source code to a path confirmed by the user.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the user's local environment, such as existing environment variables and project configuration files, to customize the generated code.
Ingestion points: Reads .env, .env.example, and project manifest files (Step 4, Step 7).
Boundary markers: The instructions specify that only missing keys should be appended to .env files and values should not be overwritten.
Capability inventory: File reading, file writing, and interactive prompting.
Sanitization: The skill uses template replacement to inject configuration into static reference code blocks.
[DYNAMIC_EXECUTION]: The skill assembles executable source code for Node.js, Python, or Java by populating reference templates with environment-specific parameters (like JWT TTL) before writing them to the disk.

jpm-oauth