# ir

> An information-retrieval substrate for agentic systems


## Pages

- [ir](index.html.md): `ir` is an information-retrieval substrate for agentic systems: one “find the relevant things in thi...
- [API reference](api.html.md): | `ir`   | `ir` — an information-retrieval substrate for agentic systems.   |
- [ir](_autosummary/ir.html.md): `ir` — an information-retrieval substrate for agentic systems.
- [ir.base](_autosummary/ir.base.html.md): Core data model for `ir`.
- [ir.cli](_autosummary/ir.cli.html.md): Command-line surface for `ir` (dispatched by cw).
- [ir.config](_autosummary/ir.config.html.md): Filesystem locations and process-wide defaults for `ir`.
- [ir.coverage](_autosummary/ir.coverage.html.md): Corpus-coverage diagnostic — what’s on disk vs. what’s actually indexed.
- [ir.embed](_autosummary/ir.embed.html.md): Embedder resolution for `ir` — a decent local default, a light fallback.
- [ir.eval](_autosummary/ir.eval.html.md): Capability-discovery evaluation — measuring how well a corpus is *found*.
- [ir.eval_gen](_autosummary/ir.eval_gen.html.md): LLM-backed generation of evaluation cases for `ir.eval`.
- [ir.formulate](_autosummary/ir.formulate.html.md): Query formulation — the Formulator seam (ir_09 §3).
- [ir.graph](_autosummary/ir.graph.html.md): The semantic link graph — typed edges between artifacts (report 12).
- [ir.index](_autosummary/ir.index.html.md): The indexing pipeline and incremental maintenance.
- [ir.maintenance](_autosummary/ir.maintenance.html.md): Idempotent background-work runner — `ir maintain` (issue #58).
- [ir.policy](_autosummary/ir.policy.html.md): Per-corpus policy — how a corpus is segmented/stored and what background work it gets.
- [ir.registry](_autosummary/ir.registry.html.md): Named-corpus registry — persistent, reusable corpus definitions.
- [ir.requirements](_autosummary/ir.requirements.html.md): Dependency check for `ir`: name what is missing and how to install it.
- [ir.rerank](_autosummary/ir.rerank.html.md): The precision stage: a cross-encoder rerank over the first-stage candidates.
- [ir.retrieve](_autosummary/ir.retrieve.html.md): Retrieval — hard metadata filtering + dense / lexical / hybrid ranking.
- [ir.schedule](_autosummary/ir.schedule.html.md): Install and operate the OS job that runs `ir maintain` — `ir schedule` (issue #75).
- [ir.sources](_autosummary/ir.sources.html.md): Defining a corpus source — an abstract strategy plus parameters.
- [ir.store](_autosummary/ir.store.html.md): Persistence for `ir` — the repository layer over `dol` key-value views.
- [ir.strategy](_autosummary/ir.strategy.html.md): Indexing strategies — the “what do we index?” seam.
- [ir.synopsis](_autosummary/ir.synopsis.html.md): Synopsis surfaces — LLM-derived summaries as an indexed surface (report 12).
- [ir.tools](_autosummary/ir.tools.html.md): Agent-callable tool surface over `ir` — plain functions returning JSON-ready
- [For AI agents](ai-agents.html.md): `ir` ships artifacts for coding agents alongside its code. This page lists
- [About this build](about-this-build.html.md): This documentation was built on **2026-10-14 21:13 UTC** from commit <a href="https://github.com/i2m...

## Optional

- [ir.md](ir.md): Complete documentation in a single file.

## Build

- [build_info.json](build_info.json): build provenance (commit, version, tool versions, whether the docs match the package)