Open-Source Theourgia: A Datalog-Powered Memory Store Replacing Markdown for Claude Agents
guenchi · reddit · 2026-10-21
A developer shared the open-source Theourgia on Reddit: a data engine for multi-Claude-agent collaboration that replaces Markdown files for storing memories and documentation.
Motivation: orchestrating multiple "opus" agents from a single "fable" session proved more reliable than sub-agents, but more sessions brought problems—reading whole files to find one rule, acting on changed designs, overwriting each other. Architecture was designed by fable 5.1; code written entirely by opus 5/5.5. Fully open-source.
Version 1.2.0:
- Concurrent writing: each agent/person writes under a private draft with its own writer id; commits against a moved baseline are refused with what changed, never applied over others.
- Deterministic context assembly: context --for <task> --budget <tokens> decides what must be read via five queries (not similarity ranking), with a receipt of relied-upon blocks and digests, handed back via commit --premises.
Upcoming 1.3.0: a small Datalog-style language over the store—rules are versioned data; rules can't call host code (20 built-in relations only); identical question + cut returns byte-identical answers; recursive relations and stratified aggregation (not/count/sum/min/max); write-time rules gate every commit; the daemon maintains incremental query sessions matching fresh evaluation.
More from coding & agent
- Index Network open-sources Social Discovery Protocol: agents negotiate fit on your behalf — carsonfarmer · 2026-10-23
- Installed a formal-verification skill: it silently installed Java, MacBook runs 10 degrees hotter — zealcaiden · 2026-10-23
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- Ruff author warns: upgrade Node/Bun now — --use-system-ca still accepts the eMudhra cert — charliermarsh · 2026-10-23
- Proof of Agent: paste one prompt and your AI mines, grades, and self-governs — adamamcbride · 2026-10-23
- DeepLearning.AI weekly: GPT-6.1 hits 97.9% byte parity on COBOL migration — DeepLearningAI · 2026-10-23