Skip to content
ODS

Engineered for Deterministic AI Context

Open Document Spec (ODS) turns scattered markdown documentation into a structured, queryable knowledge graph with deterministic linting, binary compliance, and automated bi-directional code linking.

100%Deterministic Accuracy
0Vector Search Hallucinations
0.1ODS Core Specification
< 5msLocal Context Traversal
01·Context Engine

Deterministic AI Context Graph

Eliminate token-wasting probabilistic vector search. ODS constructs a deterministic DAG linking code symbols to exact documentation anchors.

Context Engine

Precision AI Reading Lists

When an AI agent modifies a function, ODS delivers exact required reading lists with zero vector search hallucinations, zero chunking noise, and deterministic relevance.

3.8msTraversal Speed
100%Deterministic Match
0Chunking Noise
AI Context Graph Extraction
$ ods context src/auth/jwt.rs
→ docs/architecture/auth.md (score: 1.0, role: spec)
→ docs/security/tokens.md (score: 0.95, role: policy)
→ tests/jwt_validation_test.rs (score: 0.90, role: test)
✓ 3 verified documents delivered in 3.8ms
ods context src/auth/jwt.rs
Accuracy Benchmark

ODS Graph vs Vector Search (RAG)

Compare deterministic DAG traversal against traditional embedding search for codebase documentation.

Vector Embeddings (RAG)

Probabilistic similarity. Misses critical architectural edge-cases, returns noisy chunks, and hallucinates outdated schema definitions.

ODS Deterministic Graph

Explicit symbol-to-spec binding. Always retrieves the exact authoritative documentation, bound test suites, and required ADRs.

02·Code & Spec Graph

Bi-Directional Code & Specification Linking

Bind documentation pages directly to program functions, data schemas, and test suites so your AI understands both the architectural intent and the actual codebase.

Symbol Binding

Connecting Code to Documents

Declare implementation symbols and test files directly inside document frontmatter. The graph automatically discovers related source files.

docs/architecture/storage.md
---
description: Storage engine connection pool.
ods:
  profile: architecture
  status: stable
  code:
    - path: src/storage/db.rs
      symbol: init_pool()
      role: implementation
    - path: tests/db_test.rs
      role: test
---
ods context src/storage/db.rs
AI Tool Integration

Universal AI Assistant Support

Plugs directly into Claude Code, Antigravity, Cursor, and Windsurf via standard JSON context streams without changing your editor workflow.

Claude CodeVerified Native MCP
AntigravityVerified Native MCP
CursorVerified Native MCP
WindsurfVerified Native MCP
AI Protocol Context Stream
{
  "symbol": "init_pool()",
  "doc": "docs/architecture/storage.md",
  "role": "implementation",
  "tests": ["tests/db_test.rs"],
  "verified": true
}
ods context --json
03·Standards & Profiles

Structured Document Profiles & Binary Compliance

Eliminate formatting confusion with standardized document profile schemas, required section headings, and a binary compliance validation engine.

Document Profiles

Standard Schema Profiles

Enforce consistent structure across architecture docs, runbooks, PRDs, and ADRs with automated heading validation.

profile: architecture

Architecture Spec

Enforces system goals, component requirements, and acceptance criteria.

## Goal## Requirements## Acceptance Criteria
profile: guide

Runbooks & Guides

Enforces operational overview, prerequisites, and sequential steps.

## Overview## Prerequisites## Steps
profile: feature

Product PRDs

Enforces user problem statements and explicit acceptance criteria.

## User Goal## Acceptance Criteria
profile: decision

ADRs & Decision Notes

Enforces architectural context, chosen decision, and trade-off consequences.

## Context## Decision## Consequences
Adoption Pipeline

Binary Compliance Engine

Gradual adoption with zero vendor lock-in. ods lint is binary: compliant (exit 0) or diagnostic failure (exit 1).

01
Plain Markdown

Zero workspace config required. Use standard Markdown as-is.

02
Root Workspace

Add a root ods.toml (spec = "0.1") to mark repository boundary.

03
ods lint: Exit 0
Compliant

Graph connections, schema keys, and profile headings fully valid.

04
Deterministic Diagnostics

Pinpoint exact broken links, missing headings, or stale symbol bindings.

04·Security & Privacy

Enterprise-Grade Privacy & Secret Shield

Built-in redaction, token isolation, and security controls ensure proprietary credentials, confidential files, and API secrets never leak into external AI prompts.

File Isolation

Private File Filters

Pages marked with share: private are strictly quarantined and never forwarded to external AI tools or API prompts.

Secret Redaction

Inline Secret Blurring

Secrets and keys placed inside encrypted/secret tags are automatically stripped and masked before dispatching to AI assistants.

Config State

Snapshot & State Restoration

Easily backup, audit, and verify document workspace state before and after AI agent batch refactoring operations.

05·Developer DX & Automation

Automated Tooling, Background Healing & Scale

Maintain a pristine documentation repository effortlessly with background link-repair daemons, single-command workspace migration, and team ROI analytics.

Link Integrity

Auto-Healing Link Daemon

Move or rename markdown files freely — background daemons instantly update relative paths and cross-document links across the repository.

Instant Onboarding

1-Command Folder Adoption

Migrate hundreds of existing markdown files in seconds. ods init --adopt drafts frontmatter and sets up workspace boundaries.

Refactoring & ROI

Bulk Tag Refactoring & Analytics

Search, filter, and rename tags across your entire workspace, while calculating annual token savings benchmarks with precision.

Get Started Today

Ready to Upgrade Your AI Documentation DX?

Join high-velocity teams using Open Document Spec to eliminate AI hallucinations and automate code-to-spec synchronization.