The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill references the use of MERGIFY_TOKEN and GITHUB_TOKEN for authentication with the Mergify API and GitHub. It correctly instructs users to provide these via environment variables and CI secret stores, which is the standard and secure method for handling sensitive tokens in automated pipelines.
[DATA_EXFILTRATION]: The junit-process and scopes-send commands transmit test reports and metadata to api.mergify.com. As this is the official domain for the skill author (mergifyio) and the primary stated purpose of the tool, this data transfer is considered safe and intended behavior.
[COMMAND_EXECUTION]: The skill provides various CLI usage examples for the mergify tool. This includes an example of using eval to process tool output in shell environments; the documentation notes that the tool provides POSIX-safe quoting for these outputs to mitigate risks of command injection.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing external JUnit XML files. These files are handled by the CLI tool's internal parser rather than being interpreted as natural language instructions by the AI, which limits the potential for data-based prompt injection attacks.

mergify-ci