A set of Python tools built on Anthropic's Claude API for content research, drafting, and event content workflows. Built by someone who uses Claude daily and thinks seriously about what these tools can and cannot do.
These are practical tools I built while teaching myself Python and the Claude API. They reflect the workflows I actually use — researching topics for essays, preparing speaker briefings, drafting event content, and transforming raw research into structured deliverables. Each tool is a small demonstration of how Claude can be directed with specificity and care to produce work that is genuinely useful rather than generically fluent.
Takes a topic and a set of source materials (URLs, PDFs, text files) and produces a structured research brief. Designed for the kind of preparation work that event content managers and editorial teams do before drafting keynotes, session descriptions, or speaker talking points.
python research_synthesizer.py --topic "AI safety and interpretability" --sources sources.txt --output brief.mdGenerates speaker preparation materials from a set of inputs: speaker bio, event theme, audience profile, session objectives. Outputs a briefing document, suggested talking points, and audience-appropriate framing notes.
python speaker_brief_generator.py --speaker "speaker_bio.txt" --event "event_details.json" --output brief.mdAn experiment in teaching Claude to write in a specific voice. Takes a set of writing samples as calibration input and produces drafts that attempt to match the rhythm, diction, and philosophical orientation of the source material. Built because I wanted to understand the boundaries of what style transfer can and cannot preserve — and what it reveals about the difference between voice and fluency.
python voice_calibrator.py --samples samples/ --prompt "Write about the democratization of intelligence" --output draft.md- Built with Python 3.11+
- Uses the Anthropic Python SDK (
anthropic) - Claude 3.5 Sonnet / Claude 3 Opus for different use cases
- Structured output via system prompts and careful prompt engineering
- No frameworks or abstractions — plain Python, readable code
pip install anthropic
export ANTHROPIC_API_KEY="your-key-here"I built these tools because I believe the most interesting work in AI right now is happening at the boundary between human judgment and machine capability. Claude is extraordinarily powerful when directed with precision and care. It is dangerously fluent when left to its own defaults. The difference is the human behind the prompt — the person who knows what they are looking for, who can tell the difference between a correct answer and a true one, and who is willing to iterate until the output meets a standard that the machine cannot set for itself.
These tools are opinionated. They encode assumptions about what good content looks like, what a useful briefing document contains, what makes a speaker preparation packet genuinely helpful rather than generically comprehensive. Those assumptions come from experience — from building event programming, writing for institutional audiences, and thinking about what it means to communicate complex ideas with both rigor and accessibility.
Built by Ethan Garofalo — writer, event content strategist, and builder at the intersection of philosophy and AI.