The Agent Skills Directory

[SAFE]: The skill is a legitimate technical resource for the Celigo platform, documenting the use of the celigo CLI and the associated JSON/YAML schemas without any malicious behaviors.
[COMMAND_EXECUTION]: The skill documents the use of the celigo CLI for creating, updating, and testing tools. These commands are standard operations for the platform and are intended for use by developers and authorized agents within the Celigo environment.
[INDIRECT_PROMPT_INJECTION]: The skill defines a framework for processing external data through tools, creating a potential surface for indirect prompt injection. However, the documentation emphasizes structured data validation and transformation. 1. Ingestion points: Tools receive data validated against an input.schema defined in SKILL.md and references/schemas/input.yml. 2. Boundary markers: The system uses JSON Schema for structural validation, though specific delimiters for natural language data are not explicitly mandated. 3. Capability inventory: Tools can perform network operations via exports, system writes via imports, and custom logic execution via script hooks referenced by _scriptId in router.yml and output.yml. 4. Sanitization: Inputs and outputs are shaped and validated via input.schema, transform rules, and mappings.

building-tools