The Agent Skills Directory

[NO_CODE] (SAFE): The skill is composed entirely of markdown instructions without any associated scripts (Python, JavaScript, etc.) or binary files. It provides logic for the AI to follow rather than code to execute.- [PROMPT_INJECTION] (SAFE): Analysis of the instructional text found no malicious override patterns, jailbreak attempts, or instructions to ignore safety guidelines. The use of restrictive language (e.g., "Do NOT allow") is strictly for workflow enforcement and does not constitute an adversarial injection.- [DATA_EXFILTRATION] (SAFE): There are no network calls, hardcoded credentials, or references to sensitive file paths. The skill operates entirely within the agent's internal reasoning context.- [INDIRECT_PROMPT_INJECTION] (LOW): While the skill ingests data from other skills like 'brainstorming', its capabilities are limited to routing and decision-making within the agent's logic flow. It lacks the execution capabilities (like shell access) that would make it a high-risk target for indirect injection.

design-orchestration