What problem does it solve?
This skill addresses the challenge of high token usage in AI agents by demonstrating a "thin router" pattern. It proves that documentation can be loaded on-demand, significantly reducing context window size and processing costs compared to automatically injecting full documentation.
Core Features & Use Cases
- Thin Router Pattern: Provides a proof-of-concept for routing tasks to either the skill itself (for simple, focused tasks) or a full agent (for complex, multi-step work), optimizing resource allocation.
- On-Demand Documentation: Utilizes
doc-lookup to retrieve only the necessary documentation sections for CLI commands, FastAPI server, database, and MCP integration, avoiding large context windows and irrelevant information.
- Token Efficiency Analysis: Explicitly highlights the substantial token savings (up to 90% reduction) achieved by using this progressive disclosure model compared to traditional automatic documentation injection.
- Use Case: If you need to understand "what columns are in the executions table," this skill demonstrates how to use
doc-lookup to retrieve just that specific database schema section, rather than loading the entire backend documentation, saving tokens and processing time.
Quick Start
Example: Understanding CLI Flags
User request: "How do I add a --json flag to dipeo run?"
Router workflow:
1. Assess: Simple task, need CLI context
2. Load doc-lookup for CLI commands section:
python .claude/skills/doc-lookup/scripts/section_search.py
--query "cli-commands"
--paths docs/agents/backend-development.md
--top 1
