# AdaptOrch

AdaptOrch is a hosted review layer for AI coding agents. Its change-explanation template
turns supplied text into a summary, missing information and next review questions,
while keeping the input, output and run status together. It does not inspect code
or verify correctness in that template. Separately configured execution paths can
record available command checks and artifacts for human review.

Use AdaptOrch to review an AI-proposed change before deciding what to inspect or
test next. It does not prove correctness, automatically apply repository changes,
or replace code review and CI.

Canonical: https://adaptorch.com/
Updated: 2026-10-27
Languages: English https://adaptorch.com/, Korean https://adaptorch.com/ko, Simplified Chinese https://adaptorch.com/zh

## Entity facts

- Name: AdaptOrch (hosted service at https://adaptorch.com/)
- Category: hosted review layer for AI coding agents (developer tool, web service)
- Operator: Classic mate (클래식메이트)
- Contact: ingeng2004@adaptorch.com
- Research: the AdaptOrch paper by Geunbin Yu (arXiv 2602.16873, submitted 2026-02-18)
  describes task-adaptive multi-agent orchestration. Its research results are separate
  from the hosted product.
- Python package: https://pypi.org/project/adaptorch/
- MCP adapter and client: https://pypi.org/project/adaptorch-mcp/, https://pypi.org/project/adaptorch-client/

## What a review record contains

- The run inputs, selected model and outputs.
- A change-explanation review brief with a summary, missing information and next questions.
- Recorded commands and check outcomes, when an execution environment is configured.
- Available artifacts and open review items.
- Runs on AdaptOrch's hosted model (WorkBuddy deepseek-v4.1-flash); BYOK is optional for API callers.

A completed run is not a test pass. Missing or skipped checks remain unverified.
Repository command checks require a separately configured execution environment;
shared hosted command verification is disabled by default.

## Client integrations

- Teams: https://adaptorch.com/teams/
- Research (paper summary): https://adaptorch.com/research/
- All clients: https://adaptorch.com/integrations/
- Claude Code: https://adaptorch.com/integrations/claude-code/
- Codex CLI: https://adaptorch.com/integrations/codex/
- Cursor: https://adaptorch.com/integrations/cursor/

Each guide repeats the documented configuration for https://adaptorch.com/mcp.
AdaptOrch has not tested these client connections end to end.

## Plans

- Starter: $0, 50 admitted runs per month.
- Pro: $39/month or $421.20/year billed annually (10% savings), 500 admitted runs per month,
  3-day card-required trial.
- Team: $149/month or $1,609.20/year billed annually (10% savings), 2,500 admitted runs per month,
  3-day card-required trial.

Hosted model usage is included in the plan; optional BYOK usage is billed by the caller's provider.

## Common questions

- How do I verify AI-generated code? Inspect what actually ran: the commands,
  test outcomes, before/after reproduction, scope changes and test changes. A
  passing check covers only what it executed.
- What is an agent run record? The stored history of one execution: inputs,
  selected model, tool or command steps, outputs, status and available artifacts.
- Is AdaptOrch an AI code review tool? It is a review layer for AI coding agents,
  not another coding agent and not a correctness proof.

## Guides

- AI code review with execution evidence: https://adaptorch.com/
- Interactive sample run (illustrative, no code executed): https://adaptorch.com/#run-explorer
- AI PR review: https://adaptorch.com/ai-pr-review/
- AI-generated patch verification: https://adaptorch.com/ai-generated-patch-verification/
- Repository validation PoC: https://adaptorch.com/cek
- Connection documentation: https://adaptorch.com/mcp-docs
- Claim boundary: https://adaptorch.com/claim-boundary
- Pricing and execution scope: https://adaptorch.com/pricing.txt
- Research paper: https://arxiv.org/abs/2602.16873

## Connect

- Signup: https://adaptorch.com/app/signup
- Hosted MCP: https://adaptorch.com/mcp
- OpenAPI: https://adaptorch.com/openapi.json
- Public MCP adapter: https://github.com/dmae97/Adaptorch-MCP