AiMatch

ADRs for AI

ADRs4AI helps people working with AI agents record project reasoning, open questions, and human rulings in a repository. Its templates and companion tools make those records searchable and reviewable across sessions.

Visit ADRs for AI Compare alternatives Own this tool? Claim it
ADRs for AI screenshot

Who it's for

  • Developers using coding agents who want project decisions and their reasoning kept alongside repository work.
  • Project leads collaborating with AI who need a human review step for AI-proposed options and recommendations.
  • Teams using GitHub repositories that want to surface and answer open questions through VS Code or the iOS client.

Not the right fit if…

  • Teams looking for transcript archival: the page explicitly says ADRs4AI is not transcript archival.
  • Teams seeking vector-database memory: the page explicitly distinguishes the method from vector-database memory.
  • Teams looking for an AI orchestration platform: the page explicitly says ADRs4AI is not one.

How it fits your workflow

Start a repository from the template, fill in the project context, and commit the initial record. Add a seed with your thinking; an AI collaborator chunks it into ADRs that preserve your words, options, reasoning, and open questions.

The AI can stake a question with options, a recommendation, rationale, confidence, and a falsifier. A person accepts, overrides, or supplies an answer; the record can be reviewed in the VS Code extension or iOS client, and agents can query it through the MCP server.

Pricing

The vendor doesn't publish prices. Check with ADRs for AI directly.

Prices checked on Oct 26, 2026 from the vendor's site. They can change; confirm before you buy.

Key features

Repository starter template
Provides ADR and seed templates, directives, an inbox protocol, and task-runner recipes. The page says teams can adopt the template as a whole or by feature.
AI-assisted decision records
An AI turns a seed into ADRs that retain the user's words, options, reasoning, and open questions. Questions can include an AI recommendation, rationale, stated confidence, and a falsifier.
Open-question tracking in VS Code
The extension provides tree views of open questions, diagnostics for malformed records as they are typed, and deep links.
MCP server for agents
The extension includes an MCP server so agents can query the decision record.
iOS review and answering
The iOS client lets users sign in, import a repository, review open questions, and accept an answer that lands as a commit.
Searchable records
Open questions can be found in the repository with grep, using the `### QST` heading pattern.

Works with

  • MCP-compatible agents
  • Claude
  • Git and GitHub repositories

Limitations to know

  • The page explicitly says ADRs4AI is not transcript archival or vector-database memory.
  • The page explicitly says ADRs4AI is not an orchestration platform.
  • The method does not claim that an AI has persistent identity; the page describes the record as the persistent element instead.

Getting started

  1. Create a project from the ADRs4AI repository template.
  2. Fill in the top block of CLAUDE.md and commit the initial record.
  3. Open Claude or another agent, share the mission, and add a first seed.
  4. Answer the first open question and locate remaining questions with the provided grep pattern.