HagiCode
HagiCode

Hagicode

Turn AI coding from open-ended chat into reviewable, reusable proposals

Every change moves through Proposal → Execution → Archive: align on the plan before touching code, then keep what you learned as team knowledge. Works with Claude Code, Codex, GitHub Copilot, and other mainstream agent CLIs.

HagiCode is a proposal-driven AI coding workbench that turns a request into a reviewable proposal, runs it through agent CLIs such as Claude Code, Codex and GitHub Copilot, and keeps the outcome as team knowledge.

RecommendedOne-time purchase · permanent license · unlocks the full feature set
Install Desktop AppWindows · macOS · Linux
Learn More

Understand HagiCode before the feature tour

Many agents, every stage, always running

Each agent takes its own task through ideate, propose, execute and archive. When one run is archived the next begins, so work keeps moving through every stage at once.

Illustrative run

Already running an agent? See what HagiCode adds on top

Pick the agent you use. Every preview is lifted from its full comparison page, nothing more.

What it is

Anthropic's agentic coding tool; one engine behind the CLI, IDE extensions, desktop app and web

Desktop and web workbench that runs Agent CLIs, Claude Code among them; not a model or an agent itself

Pricing and licensing

Needs a Pro ($20 per month), Max (from $100), Team or Enterprise plan, a Console account, or Bedrock, Vertex or Foundry; free claude.ai plan excluded (as of October 2026)

Desktop and Container editions are free; Store edition is free with an optional one-time TurboEngine purchase

Parallel work

Subagents, plus background agents to run several sessions and watch them from one screen

Several sessions side by side, each an Agent with a visible status, across any Agent CLI

Spec and planning workflow

Plan mode, CLAUDE.md memory, skills and hooks; no proposal archive of its own

OpenSpec proposals: propose, validate, apply, archive; free editions allow 3 concurrent proposals

Read the full Claude Code comparison

What it is

Terminal AI coding agent from Tencent Cloud; sibling products are the CodeBuddy IDE and editor plugins

Desktop and web workbench that runs Agent CLIs, CodeBuddy among them; not a model or an agent itself

Pricing and licensing

Free, Pro (about $10 per month) and Team (about $40 per seat) plans, credit-based; as of October 2026, per Tencent Cloud's billing page

Desktop and Container editions are free; Store edition is free with an optional one-time TurboEngine purchase

Parallel work

Sub-agents inside a session, and Agent Teams exposed over ACP

Several sessions side by side, each an Agent with a visible status

Spec and planning workflow

Plan permission mode, /init for project context, hooks and skills

OpenSpec proposals: propose, validate, apply, archive; free editions allow 3 concurrent proposals

Read the full CodeBuddy comparison

What it is

OpenAI's terminal coding agent; the same product also has an IDE extension, a desktop app, a web version and cloud tasks

Desktop and web workbench that runs Agent CLIs, Codex among them; not a model or an agent itself

Pricing and licensing

CLI repository is Apache-2.0; usage comes with ChatGPT plans (Free, Go $8, Plus $20, Pro from $100 per month, as of October 2026) or API billing

Desktop and Container editions are free; Store edition is free with an optional one-time TurboEngine purchase

Parallel work

Subagents inside a session; codex cloud moves work to OpenAI's cloud

Several sessions side by side, each an Agent with a visible status

Spec and planning workflow

AGENTS.md via /init, permission levels via /permissions; no proposal archive of its own

OpenSpec proposals: propose, validate, apply, archive; free editions allow 3 concurrent proposals

Read the full Codex comparison

What it is

Terminal AI coding agent from GitHub, part of the Copilot product line

Desktop and web AI programming workbench that drives Agent CLIs you install

Pricing and licensing

Needs a Copilot plan: Free tier, or paid Pro (from $10/month as of October 2026), Pro+, Max, Business, Enterprise

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

One conversation per terminal; delegates to built-in and custom sub-agents

Several concurrent sessions across repos, managed from an Agents view

Spec and planning workflow

Plan mode, toggled with Shift+Tab inside a session

OpenSpec proposals: propose, validate, apply, archive; 3 concurrent on free editions, 32 with TurboEngine

Read the full GitHub Copilot comparison

What it is

LangChain's open-source Deep Agents harness; the terminal coding agent is Deep Agents Code (dcode)

Workbench that runs agent CLIs, DeepAgents included, as executors

Pricing and licensing

MIT-licensed and free; you pay your model provider; LangSmith tracing is a separate optional service

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

Sub-agents with isolated context inside a session; sessions can be resumed

Several sessions at once, each a named Agent with visible status

Spec and planning workflow

The agent plans and manages its own context; no proposal lifecycle

OpenSpec proposals: propose, validate, apply, archive (3 concurrent free, 32 with TurboEngine)

Read the full DeepAgents comparison

What it is

Open-source agent harness from DeepSeek AI, built from plugins on the Cordis framework; developer preview

Workbench that runs agent CLIs, dsh included, as executors

Pricing and licensing

MIT-licensed and free to run; you pay your model provider

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

Sessions in the Web UI; the agent can delegate work inside a session

Several sessions at once, each a named Agent with visible status

Spec and planning workflow

The agent maintains a plan inside a session; no proposal lifecycle

OpenSpec proposals: propose, validate, apply, archive (3 concurrent free, 32 with TurboEngine)

Read the full DeepSeek Harness (dsh) comparison

What it is

Open-source terminal AI agent from Google for Gemini models

Desktop and web AI programming workbench that drives Agent CLIs you install

Pricing and licensing

Apache 2.0 licence; model use needs a paid Gemini API key, Vertex/Enterprise Agent Platform, or a Code Assist Standard/Enterprise licence

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

One session per terminal; subagents inside a session

Several concurrent sessions across repos, managed from an Agents view

Spec and planning workflow

Plan mode, todo-based task planning, checkpointing and rewind

OpenSpec proposals: propose, validate, apply, archive; 3 concurrent on free editions, 32 with TurboEngine

Read the full Gemini comparison

What it is

Open-source self-improving agent with memory, skills, scheduler and chat gateway; coding is one use

Workbench that runs agent CLIs, Hermes included, as coding executors

Pricing and licensing

MIT-licensed and free; you pay your model provider, or an optional Nous Portal subscription

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

Spawns isolated sub-agents; seven terminal backends including Docker, SSH, Modal and Daytona

Several sessions at once, each a named Agent with visible status

Spec and planning workflow

No proposal workflow; learns reusable skills from experience

OpenSpec proposals: propose, validate, apply, archive (3 concurrent free, 32 with TurboEngine)

Read the full Hermes Agent comparison

What it is

JetBrains' LLM-agnostic coding agent for the terminal, IDE and CI

Workbench that runs agent CLIs, Junie included, as executors

Pricing and licensing

Billed through a JetBrains plan (free Junie Lite model, AI Pro, AI Ultimate credits), a Junie API key or your own keys; see JetBrains' pricing page

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase; model cost stays with Junie

Parallel work

One session per terminal or CI job; you coordinate parallel runs yourself

Several sessions at once, each a named Agent with visible status

Spec and planning workflow

Plan mode writes a design document from read-only analysis

OpenSpec proposals: propose, validate, apply, archive (3 concurrent free, 32 with TurboEngine)

Read the full Junie comparison

What it is

Moonshot AI's terminal coding agent; the original Kimi is archived and succeeded by Kimi Code (MIT)

Desktop and web AI programming workbench that drives Agent CLIs you install

Pricing and licensing

Kimi Code is in Kimi memberships from Plus up (Go excludes it), or use a Moonshot Open Platform API key; see kimi.com pricing

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

One session per terminal; built-in subagents (coder, explore, plan)

Several concurrent sessions across repos, managed from an Agents view

Spec and planning workflow

Plan subagent and plugins; no proposal lifecycle

OpenSpec proposals: propose, validate, apply, archive; 3 concurrent on free editions, 32 with TurboEngine

Read the full Kimi comparison

What it is

Terminal agent from Kiro; shares core capabilities with other Kiro surfaces

Desktop and web AI programming workbench that drives Agent CLIs you install

Pricing and licensing

Part of Kiro plans: Free (50 credits/month), Pro $20, Pro+ $40, Power $200 per month, as of October 2026

Desktop and Container free; Store edition free with optional one-time TurboEngine purchase

Parallel work

One chat per terminal; custom agents; cloud sessions keep running after you disconnect

Several concurrent sessions across repos, managed from an Agents view

Spec and planning workflow

Plan agent explores code without modifying it; steering files carry team rules

OpenSpec proposals: propose, validate, apply, archive; 3 concurrent on free editions, 32 with TurboEngine

Read the full Kiro comparison

What it is

Terminal coding agent, a fork of Pi, with LSP, debugger, browser and subagent tools built in

AI programming workbench that runs agent CLIs (including omp) as executors

Pricing and licensing

MIT-licensed and free; you pay only for the models you connect

Desktop and Container editions free; Store edition free base, optional one-time TurboEngine purchase

Parallel work

Subagents run in isolated worktrees inside one session

Several sessions side by side, managed as named Agents

Spec and planning workflow

Planning model role and a review command; proposal tracking is up to you

OpenSpec proposals: propose, validate, apply, archive

Read the full OMP comparison

What it is

Open-source AI coding agent with build and plan agents, LSP support and share links

AI programming workbench that runs agent CLIs, including OpenCode, as executors

Pricing and licensing

MIT-licensed and free; optional OpenCode Zen (pay-as-you-go) and Go ($10/month, as of October 2026) models

Desktop and Container editions free; Store edition free base, optional one-time TurboEngine purchase

Parallel work

Multiple agents can run in parallel on the same project

Several sessions side by side, managed as named Agents

Spec and planning workflow

Plan agent is read-only and suggests changes; Tab switches to build

OpenSpec proposals: propose, validate, apply, archive

Read the full OpenCode comparison

What it is

Minimal terminal agent harness, extended through extensions, skills, prompt templates and themes

AI programming workbench that runs agent CLIs, including Pi, as executors

Pricing and licensing

MIT-licensed and free; you pay only for the models you connect

Desktop and Container editions free; Store edition free base, optional one-time TurboEngine purchase

Parallel work

No built-in sub-agents; add them as an extension or package

Several sessions side by side, managed as named Agents

Spec and planning workflow

No plan mode by default; add one through extensions

OpenSpec proposals: propose, validate, apply, archive

Read the full Pi comparison

What it is

Terminal AI coding assistant from Qoder, part of a family with an IDE, a JetBrains plugin, an Agent SDK and cloud agents

Desktop and web workbench that runs Agent CLIs, including Qoder, as executors; it is not a model or an agent itself

Pricing and licensing

Free plan, then Pro, Pro+ and Ultra from $20 to $200 per month, credit-based (as of October 2026; see Qoder's pricing page for CLI terms)

Desktop and Container editions are free; Store edition is free with an optional one-time TurboEngine purchase

Parallel work

Subagents inside a session; for separate tasks you open more terminals yourself

Several sessions side by side, each shown as an Agent with a visible status

Spec and planning workflow

Plan mode and Goal mode inside the session; AGENTS.md for project memory

OpenSpec proposals: propose, validate, apply, archive; free editions allow 3 concurrent proposals

Read the full Qoder comparison

What it is

Open-source coding agent: reads code, edits files, runs commands and tests in your project folder

AI programming workbench that runs agent CLIs, including Reasonix, as executors

Pricing and licensing

MIT-licensed and free; you pay the model provider directly

Desktop and Container editions free; Store edition free base, optional one-time TurboEngine purchase

Parallel work

Subagent profiles within a session

Several sessions side by side, managed as named Agents

Spec and planning workflow

Plan mode: work starts only after you approve the plan

OpenSpec proposals: propose, validate, apply, archive

Read the full Reasonix comparison

Every change follows the same three steps

No open-ended chat threads. Each change becomes a proposal you can review, a run you can trace, and knowledge you can reuse.

  1. Proposal

    Describe the change. HagiCode checks your repository and drafts a proposal with specs and tasks you can read and correct before any code is touched.

  2. Execution

    Approve the plan and agents carry it out task by task, in parallel across Claude Code, Codex, GitHub Copilot and more, ticking each task off as it lands.

  3. Archive

    Finished changes are archived back into your specs. The next proposal reads them first, so it starts from what the team already learned.

Built for real development work

Proposals keep every change reviewable; parallel agents keep the work moving

Smart

Propose first, execute second: every change leaves a trail

What each change leaves behind01 / 08 · Review
  • openspec/specs/
  • openspec/changes/<change-id>/
  • proposal.md
  • design.md
  • tasks.md
  • specs/<capability>/spec.md
  • openspec validate --strict
  • openspec/changes/archive/

HagiCode doesn't just chat about your code. It reviews the real project state, writes a proposal and spec deltas, validates them strictly, runs the tasks in order, then archives what it learned for the next change.

Efficient

Run many agents and instances in parallel so your quota never sits idle

One at a time, waiting in line
Side by side, none waiting on another

Claude Code, Codex, GitHub Copilot, OpenCode, Hermes and the other supported Agent CLIs can each run several instances at once. Proposals, fixes and reviews get their own lane, so one slow result never stalls the rest.

Parallel execution matrix
Example setup
01
Claude Code
Proposal / review
01Draft spec
02Check design
03Archive review
02
Codex
Implementation / fixes
01Build feature A
02Build feature B
03Regression fix
03
Qwen · GLM
Analysis / support
01Trace issue
02Polish copy
03Capture insights
Illustrative: every lane keeps moving, none blocked by another
Agent CLIs you can equipPick one to see how it plugs in
  1. Agent
  2. Claude Code
  3. Compatible model
What it is
Anthropic's agentic coding tool; one engine behind the CLI, IDE extensions, desktop app and web
Models and sign-in
Claude models through your Anthropic plan or a third-party cloud provider
Where it runs
Terminal CLI on macOS, Windows (native or WSL) and Linux; VS Code, JetBrains, desktop app and claude.ai/code
Getting started
One native installer, Homebrew, WinGet or npm command, then browser sign-in
Known for
  • Many surfaces, one engine
  • Install paths and sign-in
  • Customisation that lives in the repo
  • Parallelism, automation and an SDK
Read the full comparison

Interesting

Every agent gets a face, a personality and a world to live in

Give each agent an avatar, Trait and Soul, then choose the vocabulary your team speaks. The same lineup reads as heroes and dungeons, a ranch, an imperial court or a hive, and levels and trophies reward the work as it happens.

One lineup, ten ways to name it
Agent
Agent
Group
Working Group
Job
Job
Log
Activity History
Progress you can feel

Every 100M tokens is one level, with a milestone cue in the status bar

Daily Achievements
  1. Star
  2. Moon
  3. Sun
  4. Crown
AgentCLI Baseline
AgentCLI Baseline
Spec StrategistAgent · Claude CodeSpec review
Patch RunnerAgent · CodexPatch fixing
Prompt ArtistAgent · Qwen · GLMPrompt styling

Example agents

See HagiCode at work

Ten areas of the product, shown in the English dark interface across five dark themes: the board, proposals, agents, skills, vaults, usage, achievements, Git and scheduled plans.

01 / 10

HagiCode session board with Proposal & Review, Execution and Archived lanes beside a project quick-actions panel

Session board

Proposals, running agents and archived work share one board, so you can see what is waiting, what is executing and what is done.

Session board

Keep exploring HagiCode in real coding sessions

After the main overview, these focused demos show how HagiCode behaves in playful and model-specific scenarios without repeating the primary product introduction.

Focused demoBilibili

AI Playing Games While Coding

A playful supporting clip that shows the assistant staying lively and surprising during real coding sessions instead of feeling like a dry code generator.

Open on Bilibili
Focused demoBilibili

GPT Codex in HagiCode: Live Trial

A focused validation run that lets prospective users judge how GPT Codex behaves inside the actual product after they've seen the main localized introduction.

Open on Bilibili

Editions & Pricing

Item
Desktop
Microsoft Store
Entry pointDesktop installMicrosoft Store install
All free features included
Vault
Skills
Proposal workflow
Local achievements
All Agent CLI integrations
Speech recognition
OmniRoute integration
GitHub integration
Git management
Maximum concurrent proposals332[1]
Document themes[1]
Custom commit signature[1]
Brand customization[1]
Custom avatar uploads[1]

Proposal concurrency counting: Generating, executing, and archiving proposals all count toward the same concurrent proposal limit.

[1] Purchase TurboEngine to unlock these Microsoft Store-only capabilities permanently. TurboEngine is a one-time perpetual license in Microsoft Store (Store ID: 9NSD809W18Z6).