The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted text from sources like Pull Requests, Slack, and Jira.
Ingestion points: The skill reads PR titles, bodies, Slack/Jira posts, and READMEs via the Read tool.
Boundary markers: There are no specific delimiters defined to separate user content from agent instructions in the prompt logic.
Capability inventory: The skill is restricted to Read and Edit tools, preventing more dangerous operations like command execution.
Sanitization: The instructions do not include specific steps to sanitize or escape the processed text to prevent embedded prompt injection attacks.
[SAFE]: No malicious patterns, obfuscation, or unauthorized external communications were detected. The references to external repositories in the metadata and credits point to a well-known service (Cursor).

deslop