Skip to content

README.md

Latest commit

 

History

History
38 lines (24 loc) · 1.63 KB

README.md

File metadata and controls

38 lines (24 loc) · 1.63 KB

Examples

These examples demonstrate the three levels of the Code-First Agents spectrum, using a GitHub issue planning workflow.

The spectrum

Level Tool What it returns LLM role
1. Data get-issue-signals.ts Raw structural signals Interprets, decides
2. Classification classify-issue.ts Complexity + score Reads classification, branches
3. Instructions analyze-issue.ts Complete procedure Executes verbatim

Each level moves more decision-making from the LLM to deterministic code.

Running the tools

All tools use Bun. They fetch a GitHub issue and output JSON to stdout.

# Level 1: raw signals
bun examples/tools/get-issue-signals.ts --owner acme --repo app --issue 42

# Level 2: classification
bun examples/tools/classify-issue.ts --owner acme --repo app --issue 42

# Level 3: instructions (the tool is a prompt factory)
bun examples/tools/analyze-issue.ts --owner acme --repo app --issue 42

Set GITHUB_TOKEN in your environment if the repo is private or you're hitting rate limits.

Skills

skills/plan-issue/ is a complete SKILL.md example that consumes analyze-issue.ts at Level 3. The skill tells the LLM to run the tool and follow the instructions field verbatim. Zero LLM branching.

Notes

These examples are simplified for illustration. Real tools handle edge cases, validation, retries, and richer output structures. The patterns are what matters, not the specific implementation.