# mesa-anyjev documentation > mesa-anyjev makes MESA's ontology and schema choices declarative and calibrated: AnyJev typed decisions (choice, yes/no, score) over mesa-mcp's OBO/OLS ontology tools, with every decision recorded in a provenance sidecar next to the mesa-ducklake AVU history, and hosted Jev through MotherDuck prompt_jev as a policy-gated third provider. This documentation covers installation, configuration, the decision graph, calibration levels and the write policy, provenance, hosted providers and development. The source repository is an Open Knowledge Format (OKF v0.2) bundle: every page carries YAML frontmatter with type, provenance (generated/sources), and lifecycle (status/stale_after) fields. Full corpus for ingestion: https://idss-mesa.github.io/mesa-anyjev/llms-full.txt Every page's Markdown source (OKF frontmatter included) is served at its URL plus `index.md` — for example https://idss-mesa.github.io/mesa-anyjev/getting-started/install/index.md. Agent guide: https://idss-mesa.github.io/mesa-anyjev/about/ai-agents/ For the decisions and verified facts behind the design, read DESIGN.md and RESEARCH.md in the repository; for behaviour, run mesa-anyjev and read its provenance sidecar. ## Getting started - [Configuration](https://idss-mesa.github.io/mesa-anyjev/getting-started/configuration/): Configure mesa-anyjev with a YAML file, MESA_ANYJEV_ environment variables or flags; precedence and the mesa-mcp fallbacks. (draft) - [Gateway](https://idss-mesa.github.io/mesa-anyjev/getting-started/gateway/): Decide with carc-fast on the CARC LiteLLM gateway at L0: the tunnel, the key, what the doctor checks, and what the gateway cannot do. (draft) - [Install](https://idss-mesa.github.io/mesa-anyjev/getting-started/install/): Install mesa-anyjev with uv, choose the extras for the gateway, local weights, Claude or Postgres, and run the doctor. (draft) - [Local weights](https://idss-mesa.github.io/mesa-anyjev/getting-started/local-weights/): Run Qwen/Qwen3-8B on a CUDA host (the GB10 workstation) for hidden states and L2 heads, the composite backend that pairs it with the gateway, and the torch 2.14 Triton switch. (draft) - [Quickstart](https://idss-mesa.github.io/mesa-anyjev/getting-started/quickstart/): Annotate one dataset card end to end with the fake backend, apply the proposals to a local DuckLake, and read the provenance back. (draft) ## Concepts - [DataCite decisions](https://idss-mesa.github.io/mesa-anyjev/concepts/datacite/): The frozen DataCite vocabularies as questions: resource type as a yes/no per member, the others as a choice with a yes/no twin, and the record fragment for mesa_avu_apply_datacite. (draft) - [Decision graph](https://idss-mesa.github.io/mesa-anyjev/concepts/decision-graph/): The fixed-key questions mesa-anyjev asks from a dataset card to a written AVU, which provider answers each, and what happens below threshold. (draft) - [Hosted providers](https://idss-mesa.github.io/mesa-anyjev/concepts/hosted-providers/): Hosted Jev through MotherDuck prompt_jev as a third provider for the same questions, and the data-residency switch that keeps it off by default. (draft) - [Learning](https://idss-mesa.github.io/mesa-anyjev/concepts/learning/): Where labels come from, how L1 temperatures and L2 heads are fitted with leave-one-card-out, the guards against positives-only data, and how an artifact is promoted. (draft) - [Levels and policy](https://idss-mesa.github.io/mesa-anyjev/concepts/levels-and-policy/): What an AnyJev level and the calibration field promise, and the rules that turn a decision into auto, proposed, human or abstain. (draft) - [MCP tools](https://idss-mesa.github.io/mesa-anyjev/concepts/mcp-tools/): The mesa_decide_* tools inside mesa-mcp: annotate, apply with one question per candidate group, explain, feedback and health; how they register, and what the elicitation state carries. (draft) - [Provenance](https://idss-mesa.github.io/mesa-anyjev/concepts/provenance/): The mesa-anyjev sidecar: runs, decisions, options, groups, AVU links, human overrides and labels, and how it joins the mesa-ducklake AVU history. (draft) - [Second opinions](https://idss-mesa.github.io/mesa-anyjev/concepts/second-opinions/): Claude as a recorded second opinion on proposed terms (never a probability, never auto), the specificity step over child terms, and the planner audit. (draft) ## Bench - [Bench results](https://idss-mesa.github.io/mesa-anyjev/bench/results/): Measured raw, L0, L1 and L2 numbers per question on local Qwen3-8B weights, the CARC gateway and the synthetic backend, every cell naming its results JSON. (draft) ## Develop - [Architecture](https://idss-mesa.github.io/mesa-anyjev/develop/architecture/): The mesa-anyjev modules, the annotate and apply phases, and the boundaries with AnyJev, mesa-mcp and mesa-ducklake. (draft) - [Testing](https://idss-mesa.github.io/mesa-anyjev/develop/testing/): How to run the hermetic mesa-anyjev test suite and the opt-in engine, Postgres, end-to-end and hosted tiers. (draft) ## About - [AI agents](https://idss-mesa.github.io/mesa-anyjev/about/ai-agents/): How AI agents should read this documentation bundle: llms.txt, the markdown mirror, trust markers. (draft) - [License](https://idss-mesa.github.io/mesa-anyjev/about/license/): mesa-anyjev is MIT licensed by the Regents of the University of New Mexico; AnyJev is Apache-2.0; NEON data are CC BY 4.0. (draft) ## Meta - [Documentation update log](https://idss-mesa.github.io/mesa-anyjev/log/): dated history of changes to this bundle.