The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided inputs to drive its research and orchestration logic.
Ingestion points: Input parameters feature_description and tech_stack defined in the Input section of SKILL.md.
Boundary markers: Prompts for sub-agents (e.g., {feature_description} or {feature}) do not include specific delimiters to isolate user input from the agent's instructions.
Capability inventory: The skill triggers various research agents (git-researcher, repo-analyst, docs-researcher, etc.) and performs dynamic skill injection.
Sanitization: No explicit validation or escaping of the user-provided feature description is performed before it is passed to subsequent agents.
[DYNAMIC_EXECUTION]: The skill implements a mechanism to load and apply additional logic at runtime.
Evidence: Step 5 of the workflow iterates through coding_styles (retrieved by the toolbox-resolver) and executes them using Skill(skill: skill). This allows the agent to dynamically adopt specific coding standards based on the project context.

blueprint-research