The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a framework for multi-step processing pipelines that ingest data from potentially untrusted sources as part of its core functionality. 1. Ingestion points: Data extracted from API responses and CSV files, plus custom prompts and configuration in .claude-flow/config.json and SKILL.md. 2. Boundary markers: The documentation does not specify the use of delimiters or 'ignore' instructions for interpolated data between chain steps. 3. Capability inventory: The skill facilitates prompt chaining, data transformation, and codebase analysis using the claude-flow CLI tool. 4. Sanitization: No explicit sanitization or validation of the ingested content is described within the skill instructions.
[SAFE]: No malicious behavior, data exfiltration, or obfuscation was found. The skill instructions follow standard workflow automation patterns and use local paths for configuration and session memory.

stream-chain