Iranti is a memory infrastructure project for multi-agent AI systems, hosted on GitHub (https://github.com/nfemmanuel/iranti）。其定位可概括为：为需要“停止遗忘”的智能体提供共享、持久、一致的知识。
Purpose According to the repository description, Iranti primarily addresses the problem of missing or inconsistent memory in multi-agent collaboration scenarios, enabling agents to share and continuously retain knowledge. It is explicitly labeled as persistent shared memory for coding agents.
Main Capabilities The capabilities directly listed in the materials include:
- Facts are stored as entity / key / value triples;
- Hybrid semantic search;
- Task checkpoints;
- Conflict resolution.
Usage Existing materials do not provide installation commands, configuration methods, interface descriptions, or usage steps; the relevant content is not yet provided.
Applicable Scenarios From the description, it is suitable for multi-agent AI systems that require shared, persistent, and consistent knowledge, especially coding agent scenarios.
Limitations The materials are only repository title and summary-type text, and do not state supported languages, frameworks, runtime environments, dependencies, license, price, or performance data; this information is not yet provided.
