The Agent Skills Directory

[SAFE]: The skill incorporates a mandatory warning and confirmation policy for track replacement operations to prevent accidental loss of user data.
[SAFE]: Explicit instructions prevent the persistence of authentication materials such as tokens, signed URLs, and cookies, effectively mitigating risks of credential exposure.
[SAFE]: The skill follows the principle of least privilege by delegating paid execution and spend approval to a separate, specialized module.
[INDIRECT_PROMPT_INJECTION]: The skill provides an interface for building CLI commands from user-supplied inputs such as titles, tags, and clip options.
Ingestion points: Metadata fields and resource identifiers within SKILL.md and references/timeline-contract.md.
Boundary markers: Present; the skill enforces a warning checklist and requires explicit user consent before executing destructive commands.
Capability inventory: Generation and execution of commands for the dreamina-canvas CLI tool.
Sanitization: The instructions provide clear formatting rules for identifiers and prohibit the use of specific prefixes to ensure CLI compatibility and reduce injection surface.

dreamina-canvas-manage-timeline