The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The SKILL.md file contains instructions for the agent to ask the user for a GAMMA_API_KEY and store it using export GAMMA_API_KEY="sk-gamma-xxxxx". While this is a common development practice, it involves the agent handling and setting sensitive environment variables.
[COMMAND_EXECUTION]: The skill uses Python scripts (scripts/generate_presentation.py and scripts/list_resources.py) to interact with the Gamma API. These scripts are executed via the shell, which is an expected part of the skill's functionality but involves subprocess execution.
[DATA_EXFILTRATION]: The skill reads the GAMMA_API_KEY from the environment and sends it as a header (X-API-KEY) to https://public-api.gamma.app. This is the intended behavior for interacting with the Gamma service, but it constitutes sending sensitive data to a remote endpoint.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to take user-provided or external text (e.g., from a GitHub repo) and send it to the Gamma API for presentation generation.
Ingestion points: The inputText parameter in the JSON payload sent to scripts/generate_presentation.py can contain untrusted data from the agent's context (e.g., SKILL.md examples show summarizing GitHub repos).
Boundary markers: None identified in the prompt interpolation logic within the provided files.
Capability inventory: The skill can perform network operations (API calls) and read environment variables.
Sanitization: There is no explicit sanitization or filtering of the inputText before it is sent to the Gamma API.

gamma