The Agent Skills Directory

[SAFE]: The skill is designed to improve the security posture of Salesforce integrations. It contains no evidence of malicious code, obfuscation, or unauthorized exfiltration. It explicitly warns against anti-patterns like committing consumer secrets to source control.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted metadata from the local project to perform security reviews, which presents a surface for indirect prompt injection.
Ingestion points: Uses Glob and Read tools to access .connectedApp-meta.xml and .eca-meta.xml files in the project directory.
Boundary markers: No explicit delimiters are used when processing metadata contents in the instructions.
Capability inventory: The skill utilizes Write, Edit, and Bash tools to manage metadata files.
Sanitization: No programmatic sanitization is applied to the ingested metadata; the skill relies on the agent's reasoning to validate configuration values against defined security scoring criteria.
[EXTERNAL_DOWNLOADS]: The documentation and asset files reference external resources for configuration and validation.
Evidence: The skill references official documentation and guidelines from trusted and well-known domains including salesforce.com, trailhead.com, github.com, and reputable community resources like salesforceben.com and apexhours.com.

sf-connected-apps