The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to process untrusted data (research notes and extracted data from external papers) by interpolating it directly into the prompt. The absence of explicit delimiters or sanitization instructions for these inputs creates an attack surface where instructions embedded in the source material could potentially influence the agent's behavior.
Ingestion points: accumulated_notes and extracted_data inputs in SKILL.md and prompt.md.
Boundary markers: Absent; input strings are processed without explicit delimiters or 'ignore' instructions.
Capability inventory: The prompt defines text synthesis, citation, and formatting tasks; no dangerous tools (shell access, file writing, network) are requested in the provided files.
Sanitization: No evidence of input validation or content filtering.
[NO_CODE]: The skill consists solely of markdown configuration and instructional prompts. It does not include any Python scripts, Node.js modules, shell scripts, or binaries, which eliminates the risk of direct malware execution or standard code-based vulnerabilities.

survey-synthesis