recursive-intelligence 0.1.3

Recursive prompt improvement — VSR loop with linguistic gate (ri-engine CLI)

Prompt Improvement Studio

Recursive prompt improvement — Variation → Selection → Retention, with a linguistic gate.

Give it a seed prompt and a goal. It runs the VSR loop offline (no API key by default), resolves language/register leanings (plain, latinate, mixed, …), and returns an improved prompt.

pip install recursive-intelligence
ri-engine improve --seed "You are a helper." --goal "When this works, the AI will resolve the task in one pass"

License: MIT · Changelog: CHANGELOG.md · Architecture: docs/technical_reference.md

---

Install

pip install recursive-intelligence
# optional real LLM backends:
pip install "recursive-intelligence[all]"

From source: git clone … && pip install -e ".[all]"

---

Usage

Primary path (any prompt, any goal)

ri-engine improve \
  --seed "You are a helper." \
  --goal "When this works, the AI will produce a structured answer with measurable success criteria"

Python API

from ri_engine import improve

result = improve(
    seed_prompt="You are a helper.",
    objective="When this works, the AI will produce a structured answer.",
)
print(result.improved_prompt)
print(result.fitness)  # VSR fitness score

Optional: template fixtures

Templates are pre-filled seed/objective pairs used for benchmarks and linguistic-registry pooling — not required for normal use.

ri-engine templates
ri-engine improve --template code-review

Proof / benchmark

ri-engine demo

Runs six fixture scenarios; scores are from the internal structural rubric (prompt_rubric.py), not live task metrics.

Expert / research commands

ri-engine expert benchmark
ri-engine expert pool-linguistic-registry
ri-engine expert macro-registry
ri-engine improve --seed … --goal … --expert

---

How it works

Linguistic Gate → Macro Priors → [Membrane] → Variation → Selection → Retention → repeat

1. Linguistic gate — picks register leaning (plain Anglo-Saxon vs latinate, etc.) from your goal and pooled registry evidence
2. Variation — generates prompt variants (8 strategies)
3. Selection — scores candidates (clarity, utility, coherence, simplicity via Occam's razor)
4. Retention — carries winning traits forward; detects convergence

Default provider is mock (offline, deterministic). Optional: --provider openai or --provider anthropic.

---

Optional providers

export OPENAI_API_KEY=sk-...
ri-engine improve --seed prompt.txt --goal "…" --provider openai

---

Docs

Doc	Contents
technical_reference.md	Architecture, VSR, macro registry, Occam's razor
publication.md	API contract, limitations, red-team notes
getting_started.md	Step-by-step CLI walkthrough
cloud_development.md	Codespaces + Colab

---

General-purpose prompt improvement. You bring the seed and goal; the engine handles language alignment and recursive refinement.