The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill documents the use of Handlebars templates (e.g., {{{lastExportDateTime}}}) and JavaScript 'hooks' (preSavePage, postResponseMap) to perform data transformation and logic at runtime. These are standard features of the Celigo integration platform intended for legitimate data processing.
[DATA_EXPOSURE]: The schema definitions in references/schemas/webhook.yml and references/schemas/distributed.yml correctly identify sensitive fields such as bearerToken, password, key, and token. These are marked with format: password or described as sensitive credentials, aligning with best practices for secret management within configuration templates.
[INDIRECT_PROMPT_INJECTION]: The skill defines a large attack surface by enabling the ingestion of data from numerous external sources (APIs, Databases, Webhooks).
Ingestion points: Defined in references/schemas/http.yml, references/schemas/webhook.yml, and various database schemas.
Boundary markers: Not explicitly defined in these technical schemas, which focus on resource structure.
Capability inventory: Includes capabilities to write to file systems via backupPath in references/schemas/file.yml and perform database updates via once.query in references/schemas/jdbc.yml.
Sanitization: Standard escaping is handled by the platform's template engine; no specific sanitization logic is present in these static schemas. This represents a known architectural pattern for integration tools and is considered a low risk in this context.
[COMMAND_EXECUTION]: The documentation describes the use of platform-specific CLI commands (e.g., celigo exports invoke, celigo account search). These are intended for use by an agent operating within the Celigo ecosystem and do not represent unauthorized command execution.

configuring-exports