The Agent Skills Directory

[SAFE]: The skill provides a legitimate implementation of a logging utility for the Cloudflare Workers platform. The code uses standard APIs (Web Crypto, console logging, R2 storage) and official deployment tools without any signs of obfuscation, exfiltration, or unauthorized privilege escalation.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for ingesting potentially untrusted data into persistent storage, which is a common characteristic of logging systems that may be targeted by indirect prompt injection.
Ingestion points: The auditLog function in lib/audit.ts accepts a data parameter containing arbitrary fields which are serialized into logs.
Boundary markers: Absent; logs are stored in NDJSON format without specific delimiters or warnings to ignore embedded instructions.
Capability inventory: The skill includes the ability to write to R2 storage buckets via env.AUDIT_LOGS_BUCKET.put within the tail worker (workers/audit-tail/src/index.ts).
Sanitization: Absent; the logging utility performs standard JSON serialization but does not sanitize or escape the content of the data record.

audit-logging-workers