Search

Utility helpers, reusable components and storage layer for Flet 0.23.0 applications

LLM-assisted analysis of classroom and small-group transcripts. Quantitative metrics, qualitative coding, structured reports.

Utility helpers, reusable components and storage layer for Flet 0.23.0 applications

Subpackage for Arango Integration in vss_ctx_rag

Subpackage for NAT Integration in vss_ctx_rag

None

ContextWall SDK - drop-in wrapper for Anthropic and OpenAI with context firewall enforcement

Graph RAG with Advanced Integration and Learning — an open-source GraphRAG library with incremental updates, file-level provenance, and pluggable LLM/storage backends.

Python SDK for the Riff Storage API

Storage Unix Style for Developers

Auto-translate GLPI ticket followups using local Ollama LLM

Local, reproducible benchmarking of open LLMs on project management tasks

Deterministic freshness and drift linter for RAG knowledge bases: catch stale chunks, missing/future timestamps, age-threshold breaches, temporal clustering, and embed-model age before they degrade your retrieval.

Static prompt-injection scanner for RAG corpora: catches jailbreak signatures, encoded payloads, hidden instructions, and role-play inducements before they reach the LLM.

Deterministic answer-grounding check for LLM/RAG outputs: catch hallucinated numbers, dates, and quotes without spending a cent on LLM-as-judge.

Lint for RAG corpora and retrievers: catch silent bugs (duplicate chunks, embed-model truncation, eval-set leakage, dim mismatch) before you ship.

Defensive, local-first secret-leak scanner for AI/LLM API keys — finds your own leaks; never validates, pools, or stores a key.

2-bit quantization with fused Metal dequant kernels for Apple Silicon — up to 8× faster local LLM inference

An AI concept-learning coach: intuition-first teaching + spaced-repetition flashcards. Understand AND remember. Portable across Claude Code, Codex, and any agent.

v3 of the positive-routing anti-slop skill. Integrates WikiProject AI Cleanup's pattern catalog into v2's editorial layer without breaking the architectural principle.