API reference

Generated from the docstrings of the public modules. The guide explains how the parts fit together; this reference lists what each module exports and the signatures.

Module	What it holds
solvi	the catalog, questions and answers, System and Response, typed facts, storage and shadow mode
solvi.decide	the decider: typed questions about a text or a state answered by a model
solvi.calibration	reliability, expected calibration error, coverage at a target accuracy
solvi.guarantee	a calibrated threshold with a stated promise on a question (System.guarantee), a computed fact or any scalar
solvi.openset	inputs from outside the calibration set: a threshold sized for the share of them, estimated as the stream goes, and a change flag
solvi.systemone	a model behind the System One HTTP API as a decider
solvi.llm	an OpenAI-compatible chat-completions server (an LLM) as a decider
solvi.remote	the client every remote model shares: retries, token counts, one policy for HTTP errors
solvi.multi	cascade, vote and route over several models
solvi.generate	generation by an LLM: a text or validated JSON, quotes checked in a text, recorded as a model's output
solvi.agree	agreement of generated candidates under a key: the chosen one, its share as a fact, the tally
solvi.refine	a check that says why (Fail); the loop propose → check → re-ask with the reasons → escalate
solvi.memory	a memory of corrected cases: nearest neighbours with an abstain threshold
solvi.lora	a LoRA adapter per question, fitted on a few hundred labelled examples
solvi.learning	learning from corrections with gates and rollback (experimental)
solvi.serve	solvi serve: HTTP, MCP and the System One API
solvi.agents	guarding an agent's tool calls: Guard, the MCP proxy (adapters for PydanticAI, LangGraph, the OpenAI Agents SDK)
solvi.hooks	a coding agent's hooks: edits checked against rules, a skill picked for a prompt, install / uninstall (Claude Code; Codex, preview)
solvi.textin	text in: a message → the question it asks and its typed input state, read with quotes
solvi.specialist	the specialist contract (preview): propose a typed spec, check it against the source, render, replay
solvi.charts	verified charts (preview): a text with numbers → an SVG in which every number is quoted from the text
solvi.longdoc	long texts: sections, BM25 retrieval, the window a decider reads
solvi.many	a choice among many options: a shortlist, then the decider
solvi.episode	an agent's memory (what was tried, what failed, what worked) as an input
solvi.worldmap	a map of an environment an agent builds by acting: claims with provenance
solvi.drift	drift: has the stream moved away from the one the thresholds were calibrated on?
solvi.heads	the learned answer heads: System.fit's FastHead and select_features, CandidateHead, Head (solvi.fast up to 0.7)
solvi.extract_long	@extract by description for long documents (LongSpanExtractor; torch)
solvi.extract_multi	a single-pass @extract over several fields (torch)
solvi.strategist	the strategist: a flow per request (Flow, PlanError), the learned order of hard checks and producer policy
solvi.strategy	the model strategist (experimental, ModelStrategist)
solvi.costs	measured run times (CostBook) and the planner's measured costs (MeasuredCosts)
solvi.rulelist	a readable rule list learned from examples (System.learn_rule; solvi.rules up to 0.7)
solvi.aliases	name matching between parameters and facts (experimental)
solvi.audit	res.audit(): what each answer rests on and which safeguards fired; to_dict() as JSON-ready data
solvi.show	printing a response
solvi.report	reports of one decision or a period of stored decisions (Markdown, HTML, data)
solvi.otel	decisions as OpenTelemetry spans (the API, or OTLP/JSON)
solvi.sets	decisions over a set: answers of many items made consistent under at-most / exactly-one / capacity constraints
solvi.search	search over alternatives: candidates through the System's checks, the best by an objective, pruned, recorded
solvi.counterfactual	counterfactuals: the smallest change of the given inputs that changes an answer
solvi.perturb	instruction-like sentences in an input and the variants without them (perturb=k)
solvi.storage	TraceStorage: JSONL, SQLite, PostgreSQL and DuckDB stores, queries, replay
solvi.signature	a signature of a trace or a store that names the one changed record (preview)
solvi.diff	solvi diff and shadow mode
solvi.testing	decision regression tests from cases.json
solvi.honesty	the honesty suite
solvi.check	solvi check: catalog lint
solvi.calibfile	calibration files (save_calibration / load_calibration) and solvi calibrate
solvi.models	solvi models: published and cached deciders, pull, check
solvi.scaffold	solvi init: a new decision project
solvi.i18n	languages of rendering: the audit, show and the safeguard report in Russian