Writing Intelligence is a living project. Here's where it's going.
Built by Antonio T. Smith Jr. / Density6 LLC
The full compiler. 52 files. 42,264 words. Everything listed in the README.
- Technical documentation
- Grant writing / NOFO responses
- Journalism / feature writing
- Resume / cover letter
- Social media (platform-specific: LinkedIn, Twitter/X, Instagram)
- Podcast show notes
- YouTube scripts
- Newsletter
- Technical explainer (clear, patient, precise)
- Investigative journalist (evidence-chasing, skeptical, specific)
- Coach / mentor (direct, warm, action-oriented)
- Satirist (dry, observational, subversive)
- Political science
- Anthropology
- Engineering
- Education / pedagogy
- Theology / religious studies
- Law (expanded beyond legal positioning)
- More before/after examples in
tests/gold_outputs.md - Expanded adversarial test cases
- Community-submitted detector benchmark results
- Spanish anti-patterns and cadence signatures
- French anti-patterns and cadence signatures
- Portuguese anti-patterns and cadence signatures
- German anti-patterns and cadence signatures
- Arabic writing conventions pack
- Mandarin writing conventions pack
- Bilingual voice handling (code-switching awareness)
- Register calibration for non-English academic conventions
- JSON schema for auto-diagnostic reports (machine-readable output)
- Scoring API specification (input text → structured score)
- Batch processing protocol for multi-document auditing
- Webhook-compatible event format for CI/CD writing checks
- Writing Intelligence as a Model Context Protocol server
- Real-time scoring endpoint
- Voiceprint matching endpoint
- Genre detection endpoint
- Automated benchmark pipeline against Turnitin, GPTZero, Originality.ai
- Regression testing: every update verified against detector pass rates
- Community leaderboard for detector benchmark results
- Shared voiceprint libraries (team voice consistency)
- Shared terminology locks (brand voice enforcement)
- Multi-author continuity tracking (books with multiple contributors)
- Statistical profiling engine (automated sentence-length, vocabulary, perplexity analysis)
- Automatic cadence signature detection (flag patterns before human review)
- Voice drift alerting (notify when output deviates from established voiceprint)
- Star the repo — signals demand and helps prioritize
- Open an issue — describe what you need and why
- Submit a PR — build it yourself and contribute back
- Share results — run the compiler on your writing, share what worked and what didn't
The roadmap is driven by community need, not a corporate product schedule. What gets built next depends on what people actually use and ask for.