The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from project files and session history to generate task handoffs and experience records, potentially propagating malicious instructions into future sessions.
Ingestion points: The handoff and reflect modules ingest content from the current session and any related project files.
Boundary markers: There are no explicit instructions or delimiters used to separate the analyzed content from the generated summaries, which could cause embedded instructions in source files to be inadvertently obeyed by the agent in future interactions.
Capability inventory: The skill performs file read and write operations across the project directory and specific absolute system paths.
Sanitization: The skill lacks a sanitization or filtering mechanism to neutralize potential prompt injection attacks hidden within the files it reads.

ylf-context-continuity