The Agent Skills Directory

[SAFE]: The skill is a documentation-only resource providing deployment best practices. It does not include scripts, command execution, or network exfiltration patterns.
[SAFE]: All external links point to the official Mistral AI documentation domain (docs.mistral.ai), which is a recognized and safe service.
[SAFE]: The instructions explicitly emphasize security controls, such as secret isolation, bounded concurrency, and the requirement for manual approval before performing sensitive deployment actions.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface by using Read, Glob, and Grep tools to inspect local code and configuration files.
Ingestion points: Tool instructions in SKILL.md for inspecting repository content.
Boundary markers: The skill includes explicit "Tool Discipline" and "Approval Boundaries" sections that limit the agent's autonomy and require human confirmation for stateful changes.
Capability inventory: The skill uses file inspection and basic editing tools (Read, Glob, Grep, Write, Edit) but lacks tools for network access, subprocess execution, or privilege escalation.
Sanitization: The skill mandates offline validation of artifacts and configuration before deployment actions are considered.

mistral-deploy-integration