v0.1.34

Added

• Fuzzy search — camelCase/snake_case tokenization, acronym matching (okf lookup repo finds UserRepository). New --exact flag for strict title-only matching.
• Dockerfile + Containerfile + docker-compose.yml parsers — FROM → docker dep with tag, RUN pip install → pip dep with version, image: and depends_on: for compose services.
• LLM enrichment CLI — okf generate --enrich activates LLM description/docstring enrichment. Gracefully skips when no API key is configured.
• .okfconfig config system — extensible flat + sectional config (global bundle_dir, feature sections llm.*, serve.*, lookup.*, mcp.*, pairs.*). No environment variables. okf config CLI to view/edit. okf init prompts for config interactively.
• Pre-commit hook — .pre-commit-config.yaml includes okf-generate hook that auto-regenerates bundle when source files change.
• Docker image — Dockerfile + GHCR publish workflow on release.
• Live Demo button on landing page with CI-published visualization.

Changed

• Config system migrated from env vars to .okfconfig file. All os.environ lookups removed from generator, pairs, serve, lookup, and init.
• Default LLM endpoint changed to http://localhost:8080/v1 (local-first, works with llama.cpp/Ollama out of the box).
• okf init wizard now prompts for config (LLM provider, model, API key) before generating bundle.
• RELEASE.md expanded with 15-point pre-release audit including feature validation and landing page freshness checks.
• README.md feature table updated with fuzzy search, Dockerfile/compose, config, pre-commit, Docker image rows.

---