The Agent Skills Directory

[DATA_EXFILTRATION]: The skill contains code templates designed to transmit error logs, performance metrics, and alerts to external service endpoints.
Evidence:
resources/implementation-playbook.md: The ErrorTracker class uses the fetch API to send JSON-encoded error events to a configured endpoint with an Authorization header.
resources/implementation-playbook.md: The SlackAlertChannel uses aiohttp to send POST requests containing alert data to a Slack webhook_url.
Note: The skill demonstrates good security practice by including a sanitizeEvent function that redacts sensitive keys such as 'password', 'token', 'secret', and 'api_key' before transmission.
[PROMPT_INJECTION]: The skill includes an automated analysis script that reads local codebase files, which creates an ingestion surface for indirect prompt injection.
Evidence Chain:
Ingestion points: resources/implementation-playbook.md contains the ErrorTrackingAnalyzer which reads the content of all files matching common programming extensions using Path(project_path).rglob('*').
Boundary markers: The provided script does not utilize boundary markers or explicit instructions to ignore embedded agent commands within the analyzed files.
Capability inventory: The skill possesses capabilities for file system traversal and network operations (HTTP POST) via the included code patterns.
Sanitization: While data sanitization is present for credential redaction in the logging service, the code analysis tool does not specifically filter or sanitize the file content it processes.

error-diagnostics-error-trace