The Agent Skills Directory

[NO_CODE]: The skill consists entirely of Markdown documentation, workflow phases, and instructional prompts. It does not contain any executable scripts (Python, JavaScript, shell) or configuration files that would perform operations on the host system.
[REMOTE_CODE_EXECUTION]: There are no patterns involving package installations (pip, npm) or remote script fetching (curl, wget). All instructions are static text meant for guidance.
[DATA_EXFILTRATION]: The skill does not access sensitive files, environment variables, or credentials. No network operations are defined.
[PROMPT_INJECTION]: The instructions use standard instructional language to guide an agent through RAG development. There are no attempts to override safety filters, bypass system instructions, or extract internal prompts.
[OBFUSCATION]: The content is provided in clear, human-readable Markdown. No Base64, hex encoding, or zero-width character obfuscation was detected.
[INDIRECT_PROMPT_INJECTION]: While the skill describes an architecture that processes user queries and external data sources (RAG), the skill itself does not provide the code to perform these operations, nor does it interpolate untrusted data into prompts in a way that creates a vulnerability surface within the skill's own scope.

rag-implementation