Grounding Python Packages

living-memory-model

A verifiable memory layer for language models — facts in a graph, every claim with its source.

groundgate

Deterministic admission for LLM-extracted facts: admit, flag for review, or reject, with verifiable receipts.

reverify

Stop your AI from making things up — it proposes, deterministic tools decide, every claim checked against ground truth with evidence. Grounded facts and context survive resets. Reverse engineering is the proving ground. MCP server + CLI.

figured

Show your work: verify that every number in an LLM-generated answer traces to the rows it was derived from. Zero dependencies.

fieldproof

Document extraction that shows its work: every LLM-extracted field grounded to its source words, verified, and reviewable.

earthing

A python library for design of earthing networks in electrical substations.

gui-agents

Agent S: an open agentic framework that uses computers like a human

sourced-evidence

Per-claim grounded/contradicted/unverified checking of LLM output against its own source context

grounder-mcp

Grounder MCP - live web search, fetch, deep_search evidence packs, and agentic research for local & cloud LLMs. Thin client for grounder.dev.

groundgate-calibrate

Deterministic admission for LLM-extracted facts: admit, flag for review, or reject, with verifiable receipts.

ancora-rag

Grounded answers, or an honest escalation. A dependency-free hallucination gate for RAG chatbots — rejects invented numbers, not just low similarity.

outcome-receipts

Beta. Funder outcome reports where every reported figure is a receipt: the exact query, the rows it drew from, a content hash of that slice, and a timestamp. Deterministic, offline, and fail-closed. Export is refused if a number in the narrative, or in a chart or comparison claim, cannot be traced to the data that produced it.

onus

A claim checker: verifies a document's claims against its evidence, numbers deterministically

bearout

A runtime grounding gate that fails closed and publishes its own error rates. Every sentence verified against its cited source, or suppressed.

verifydoc

Trust layer for document→JSON extraction & AI agents: calibrated per-field confidence + source grounding + accept/review abstention on any OCR/VLM. Ships VerifyDocBench, a novel grounding-conditioned conformal method, and an MCP server.

groundguard

Verify LLM output against your source documents. Catch hallucinations in RAG pipelines and agentic workflows before they reach users.

groundguard-ai

Deterministic fact gate for tool-using AI agents.

evidence-gate-py

Stop your AI from making up facts about your own data. A pre-generation grounding gate for LLM/RAG apps — zero-dep, MIT, JS + Python, with an MCP server.

proof-engine-wiki

AI agent skill that creates formal, verifiable proofs of claims — every fact computed or cited, never asserted

proof-citations

AI agent skill that creates formal, verifiable proofs of claims — every fact computed or cited, never asserted

proof-engine-registry

AI agent skill that creates formal, verifiable proofs of claims — every fact computed or cited, never asserted

gemini-groundcite

An Open Source Python package to strengthen Gemini Citations- Validated, Controlled, Reliable and Structured Citations

paperpin

Pin every LLM-extracted value to the exact spot on the page it came from, and flag the values that aren't there at all. CPU-only, any model, zero telemetry.

tina4-groundwire

Retrieval + verification in front of an LLM — ground a weak reader in your own source, off-GPU. Part of the tina4stack.