The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation in references/execution-modes.md includes shell commands for environment detection, such as verifying the presence of the Salesforce CLI (sf), checking for local project files (sfdx-project.json), and confirming write permissions on the filesystem.
[EXTERNAL_DOWNLOADS]: The skill describes a protocol for handling large datasets by downloading payloads from an artifactUrl provided in API responses. This is presented as a standard method for pagination and processing large volumes of Salesforce data when the payload exceeds context limits.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources for reporting purposes. 1. Ingestion points: Salesforce SOQL query results and local CSV files in the sample-data/ directory. 2. Boundary markers: None identified in the provided instructions. 3. Capability inventory: Salesforce data retrieval (soql_query), object metadata inspection (sobject_describe), shell command execution for environment checks, and document generation. 4. Sanitization: No explicit sanitization or escaping logic is defined for the interpolated record data.

sf-campaigns