The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines workflow patterns that ingest external data through an inputs dictionary, which could be exploited via indirect prompt injection if the data contains malicious instructions for the downstream tasks or LLM calls.
- Ingestion points: inputs parameter in @entrypoint decorated functions including my_workflow, research_workflow, and strands_workflow within SKILL.md.
- Capability inventory: The skill utilizes capabilities such as network requests via requests.post and call_api, web/document searches, and LLM interactions via llm.invoke across various code snippets in SKILL.md.
- Boundary markers: No delimiters or explicit instructions to ignore embedded commands are present in the provided templates.
- Sanitization: No input validation, escaping, or filtering mechanisms are implemented in the code examples.

langgraph-functional-api