fairpeer

module

v0.1.5 Latest Latest Go to latest Published: Aug 9, 2026 License: MIT

Details

Valid go.mod file
Redistributable license
Tagged version
Stable version
Learn more about best practices

Repository

github.com/zzycxz/fairpeer

Links

Open Source Insights

README ¶

fairpeer

English · 简体中文 · User Guide · Architecture Spec

Universal Multi-Vendor AI Coding Assistant + Office Automation Platform.

Connect to 18 vendors (11 direct + 7 Coding Plan aggregators), 300+ models.
Built-in Word/Excel/PPT office automation capabilities.
Single static Go binary, zero runtime dependencies, seamless cross-platform coverage.

What is fairpeer?

fairpeer is a universal AI coding assistant that supports 11 direct vendors and 7 Coding Plan aggregator platforms, providing complete office automation capabilities.

Core Features

🌐 Multi-Vendor Support — 18 vendors, unified configuration management
💻 Code Editing — 5 editing strategies + 5-level fuzzy matching + checkpoint rollback
📄 Office Automation — Full Word/Excel/PPT support
🧠 Smart Memory — Two-layer long-term memory + Dream agent + RAG knowledge base
🎯 Task Orchestration — Auto-plan + Goal mode + Expert teams
🔧 Plugin System — Markdown Skill + MCP plugins
📷 Screenshot Solver — One-hotkey exam assistant, AI reads screen and answers

🚀 Quick Start

Installation

# macOS / Linux
curl -fsSL https://fairpeer.dev/install.sh | bash

# Windows (PowerShell)
irm https://fairpeer.dev/install.ps1 | iex

# Or download from GitHub Releases
# https://github.com/zzycxz/fairpeer/releases

Configuration

# Initialize configuration
fairpeer setup

# Or manually edit config file
vim ~/.config/fairpeer/fairpeer.toml

Run

# Start interactive chat
fairpeer

# Execute a single task
fairpeer run "Help me write a Hello World"

# Start desktop app
fairpeer-desktop

🌐 Supported Vendors

Direct Vendors (11)

Vendor	Base URL	Default Model
Qwen	`https://dashscope.aliyuncs.com/compatible-mode/v1`	qwen3.7-max
DeepSeek	`https://api.deepseek.com`	deepseek-v4-pro
Volcengine	`https://ark.cn-beijing.volces.com/api/v3`	doubao-seed-evolving
Zhipu AI	`https://open.bigmodel.cn/api/paas/v4`	glm-5.2
MiniMax	`https://api.minimaxi.com/v1`	minimax-m3
Moonshot	`https://api.moonshot.cn/v1`	kimi-k3
MiMo	`https://api.xiaomimimo.com/v1`	mimo-v2.5-pro
StepFun	`https://api.stepfun.com/v1`	step-3.7-flash
iFlytek MaaS	`https://spark-api-open.xf-yun.com/v1`	glm-5.2
Anthropic	`https://api.anthropic.com`	claude-sonnet-5
OpenAI	`https://api.openai.com/v1`	gpt-5.6-terra

Coding Plan Aggregators (7)

Platform	Base URL	Description
Qwen Coding	`https://coding.dashscope.aliyuncs.com/v1`	Aggregates qwen/kimi/glm/minimax
Zhipu GLM Coding	`https://open.bigmodel.cn/api/coding/paas/v4`	GLM series
Volcengine Coding	`https://ark.cn-beijing.volces.com/api/coding/v3`	Aggregates doubao/deepseek/kimi
Baidu Qianfan	`https://qianfan.baidubce.com/v2/coding`	Aggregates ernie/glm/kimi/deepseek
Tencent TokenHub	`https://api.lkeap.cloud.tencent.com/coding/v1`	Aggregates deepseek/glm/kimi/minimax
StepFun Step Plan	`https://api.stepfun.com/step_plan/v1`	step series
iFlytek MaaS Coding	`https://maas-token-api.cn-huabei-1.xf-yun.com/v2`	Unified model ID

📄 Office Automation

Word Documents

# Create Word document
fairpeer run "Create a project report.docx with title, paragraphs and tables"

# Insert images
fairpeer run "Insert logo.png into report.docx"

# Generate table of contents
fairpeer run "Generate table of contents for report.docx"

Excel Spreadsheets

# Create Excel spreadsheet
fairpeer run "Create sales_data.xlsx with months and sales"

# Add charts
fairpeer run "Add bar chart to sales_data.xlsx"

# Add conditional formatting
fairpeer run "Add conditional formatting to sales column in sales_data.xlsx"

PowerPoint Presentations

# Create PPT
fairpeer run "Create a presentation about AI"

# Use template
fairpeer run "Create PPT using template, topic is digital transformation"

# Add animation
fairpeer run "Add fade-in animation to PPT"

🧠 Smart Memory System

Long-term Memory

FairPeer supports two-layer long-term memory:

Document Layer — fairpeer.md file, automatically loaded into context
Auto Memory — Dream agent automatically consolidates knowledge

RAG Knowledge Base

# Import documents
fairpeer run "Import this PDF into knowledge base"

# Search knowledge base
fairpeer run "Search for authentication system knowledge"

# Answer questions using knowledge base
fairpeer run "Answer this technical question using knowledge base"

🎯 Task Orchestration

Auto-plan (Automatic Planning)

# Complex tasks automatically trigger planning
fairpeer run "Refactor authentication module, add JWT support, write unit tests"
# → Auto plan → User approval → Execute

Goal (Goal Mode)

# Autonomous execution until completion
fairpeer goal "Fix all TypeScript errors"
# → Autonomous work → Detect blockers → Complete/Report

Expert Teams

# Multi-expert collaboration
fairpeer team review "Review this PR"
# → Security expert + Performance expert + Architecture expert → Comprehensive report

🔧 Plugin System

Skills

# List available skills
fairpeer skill list

# Install skill
fairpeer skill install code-review

# Create custom skill
fairpeer skill new my-skill

MCP Plugins

# fairpeer.toml
[[plugins]]
name = "my-mcp-server"
type = "stdio"
command = "node"
args = ["server.js"]

⚙️ Configuration

Config File Locations

~/.config/fairpeer/fairpeer.toml    # Global config
./fairpeer.toml                     # Project config
./.env                              # Environment variables

Configuration Example

# Default model
default_model = "deepseek/deepseek-v4-pro"

# Provider configuration
[[providers]]
name = "deepseek"
kind = "openai"
base_url = "https://api.deepseek.com"
api_key_env = "DEEPSEEK_API_KEY"
default = "deepseek-v4-pro"
fast_model = "deepseek-v4-flash"
context_window = 1000000

# Context management
[agent]
soft_compact_ratio = 0.5
compact_ratio = 0.8
compact_force_ratio = 0.9

# Office automation
[cowork]
vlm_model = "openai/gpt-4o"
screenshot_vlm_model = "openai/gpt-4o"

🛠️ Development

Build

# Clone repository
git clone https://github.com/zzycxz/fairpeer.git
cd fairpeer

# Install dependencies
go mod download

# Build CLI
go build -o fairpeer ./cmd/fairpeer

# Build desktop app
cd desktop
wails build

Test

# Run all tests
go test ./...

# Run specific tests
go test ./internal/config/...

📚 Documentation

Reference	Content
User Guide	Permissions, sandbox, MCP plugins, slash commands, `@` syntax, Plan mode
Architecture Spec	Engineering contract: system architecture, registry mechanism, data types
Features	Complete feature panorama, architecture diagram, capability matrix
Office Automation	Word/Excel/PPT automation, email integration, calendar tasks
RAG Knowledge Base	Document import, knowledge graph, entity extraction, semantic search
Expert Teams	Multi-model collaboration, team configuration, collaboration modes
Contributing	Developer guide: how to add new tools, providers, and bot channels
Changelog	Historical release records and feature iterations

💰 Sponsorship

If you find FairPeer useful, please consider sponsoring us. Your support will be used for:

🤖 Token purchases for development and testing
🚀 Feature development and maintenance
📚 Documentation improvements

PayPal

Alipay

🤝 Contributing

Contributions are welcome! Please see CONTRIBUTING.md.

📄 License

MIT License - see LICENSE

🔗 Links

FairPeer — The most powerful universal multi-vendor AI coding assistant!

Directories ¶

?	: This menu
/	: Search site
f or F	: Jump to
y or Y	: Canonical URL

Path	Synopsis
benchmarks
context-maintenance-e2e command Drives the context-maintenance E2E scenarios against a real LLM API: seed → (idle past cache TTL) → resume A/B-compares cold-restart miss tokens with and without pruning.	Drives the context-maintenance E2E scenarios against a real LLM API: seed → (idle past cache TTL) → resume A/B-compares cold-restart miss tokens with and without pruning.
cmd
cua-replay command Command cua-replay is a debugging tool for the CUA (Computer-Use Agent) vision loop.	Command cua-replay is a debugging tool for the CUA (Computer-Use Agent) vision loop.
e2ebench command e2ebench runs the committed e2e task suite against a real provider and emits a markdown + JSON report (accuracy, cache-hit rate, token use, cost) for a PR.	e2ebench runs the committed e2e task suite against a real provider and emits a markdown + JSON report (accuracy, cache-hit rate, token use, cost) for a PR.
fairpeer command Command fairpeer is a config- and plugin-driven coding agent CLI.	Command fairpeer is a config- and plugin-driven coding agent CLI.
fairpeer-plugin-example command Command fairpeer-plugin-example is a reference fairpeer plugin: a minimal MCP stdio server speaking newline-delimited JSON-RPC 2.0 on stdin/stdout.	Command fairpeer-plugin-example is a reference fairpeer plugin: a minimal MCP stdio server speaking newline-delimited JSON-RPC 2.0 on stdin/stdout.
desktop module
internal
acp Package acp implements the Agent Client Protocol (https://agentclientprotocol.com) transport: a stdio JSON-RPC 2.0 agent that editors and other host clients speak to drive fairpeer.	Package acp implements the Agent Client Protocol (https://agentclientprotocol.com) transport: a stdio JSON-RPC 2.0 agent that editors and other host clients speak to drive fairpeer.
agent Package agent wires a Provider, a tool Registry, and a Session into the harness loop that drives a coding task to completion.	Package agent wires a Provider, a tool Registry, and a Session into the harness loop that drives a coding task to completion.
agent/testutil Package testutil provides reusable test helpers for the agent package.	Package testutil provides reusable test helpers for the agent package.
apihelper Package apihelper provides a small set of shared HTTP helpers reused by the scheduler LLM time-parser, the RAG ask path, and the RAG extractor.	Package apihelper provides a small set of shared HTTP helpers reused by the scheduler LLM time-parser, the RAG ask path, and the RAG extractor.
assets Package assets embeds the built-in, cross-platform skill payloads that fairpeer ships with its binary.	Package assets embeds the built-in, cross-platform skill payloads that fairpeer ships with its binary.
boot Package boot assembles a ready-to-drive control.Controller from configuration: it loads config, resolves the model(s), builds the tool registry (built-ins + plugins), wires the permission gate, and constructs the executor.	Package boot assembles a ready-to-drive control.Controller from configuration: it loads config, resolves the model(s), builds the tool registry (built-ins + plugins), wires the permission gate, and constructs the executor.
bot Package bot 实现 fairpeer 多渠道 IM bot 消息网关，支持 QQ、飞书、微信、Telegram。	Package bot 实现 fairpeer 多渠道 IM bot 消息网关，支持 QQ、飞书、微信、Telegram。
bot/feishu Package feishu 实现飞书自建应用 Bot 适配器。	Package feishu 实现飞书自建应用 Bot 适配器。
bot/qq Package qq 实现 QQ 官方 Bot API v2 适配器。	Package qq 实现 QQ 官方 Bot API v2 适配器。
bot/telegram Package telegram 实现 Telegram Bot API 适配器。	Package telegram 实现 Telegram Bot API 适配器。
bot/weixin Package weixin 实现微信 iLink Bot 适配器。	Package weixin 实现微信 iLink Bot 适配器。
browserlaunch Package browserlaunch launches a system-installed Chromium-based browser (Chrome / Edge / Brave / Chromium) with a CDP remote-debugging endpoint and exposes the resulting websocket URL so a driver — the Python browser-use sidecar, or chromedp elsewhere — can attach to the very same browser instead of each driver spawning its own.	Package browserlaunch launches a system-installed Chromium-based browser (Chrome / Edge / Brave / Chromium) with a CDP remote-debugging endpoint and exposes the resulting websocket URL so a driver — the Python browser-use sidecar, or chromedp elsewhere — can attach to the very same browser instead of each driver spawning its own.
browseruse Package browseruse is the Go client for the Python browser-use sidecar (browseruse_server.py).	Package browseruse is the Go client for the Python browser-use sidecar (browseruse_server.py).
builtinmcp Package builtinmcp defines MCP servers that ship with fairpeer without requiring user configuration.	Package builtinmcp defines MCP servers that ship with fairpeer without requiring user configuration.
calendar Package calendar provides a local SQLite-backed calendar for FairPeer.	Package calendar provides a local SQLite-backed calendar for FairPeer.
checkpoint Package checkpoint is fairpeer's snapshot-based edit safety net.	Package checkpoint is fairpeer's snapshot-based edit safety net.
cli Package cli implements fairpeer's command-line entry: subcommand routing, flag parsing, assembly from config, and exit codes.	Package cli implements fairpeer's command-line entry: subcommand routing, flag parsing, assembly from config, and exit codes.
codegraph Package codegraph integrates the CodeGraph code-intelligence engine (https://github.com/colbymchenry/codegraph) as a built-in MCP server.	Package codegraph integrates the CodeGraph code-intelligence engine (https://github.com/colbymchenry/codegraph) as a built-in MCP server.
command Package command loads custom slash commands from Markdown files.	Package command loads custom slash commands from Markdown files.
compose Package compose implements fairpeer's structured development workflow: after a plan is approved, a multi-task feature runs through Implement → Verify → (Review) with bounded retries, instead of a single execution turn that never validates the result.	Package compose implements fairpeer's structured development workflow: after a plan is approved, a multi-task feature runs through Implement → Verify → (Review) with bounded retries, instead of a single execution turn that never validates the result.
config Package config loads fairpeer's runtime configuration from TOML.	Package config loads fairpeer's runtime configuration from TOML.
control Package control is the transport-agnostic session driver.	Package control is the transport-agnostic session driver.
diff Package diff computes a line-level diff between two versions of a file and renders it as a unified diff, with added/removed line counts.	Package diff computes a line-level diff between two versions of a file and renders it as a unified diff, with added/removed line counts.
docconv Package docconv centralizes the external Python-script document-conversion helpers (markitdown via doc_converter.py, and PDF OCR via ocr_pdf.py) that were previously duplicated across three packages (rag, tool/builtin, control).	Package docconv centralizes the external Python-script document-conversion helpers (markitdown via doc_converter.py, and PDF OCR via ocr_pdf.py) that were previously duplicated across three packages (rag, tool/builtin, control).
doctor Package doctor collects local, redacted diagnostics for issue reports.	Package doctor collects local, redacted diagnostics for issue reports.
event Package event defines the typed event stream the agent emits as it runs a turn, and the Sink it emits to.	Package event defines the typed event stream the agent emits as it runs a turn, and the Sink it emits to.
evidence
experts
fileref
fileutil
fileutil/encoding Package encoding detects and converts file encodings for the built-in file tools.	Package encoding detects and converts file encodings for the built-in file tools.
frontmatter Package frontmatter provides a minimal, dependency-free parser for the ---fenced "key: value" blocks that prefix skill, command, and memory files.	Package frontmatter provides a minimal, dependency-free parser for the ---fenced "key: value" blocks that prefix skill, command, and memory files.
hook Package hook runs user-configured shell-command hooks around the agent loop: PreToolUse / PostToolUse fire around each tool call, UserPromptSubmit before a turn, Stop after it.	Package hook runs user-configured shell-command hooks around the agent loop: PreToolUse / PostToolUse fire around each tool call, UserPromptSubmit before a turn, Stop after it.
i18n Package i18n holds the CLI's translatable strings and a small detection helper.	Package i18n holds the CLI's translatable strings and a small detection helper.
installsource Package installsource implements the `install_source` tool: a two-phase installer for fairpeer skills and MCP servers.	Package installsource implements the `install_source` tool: a two-phase installer for fairpeer skills and MCP servers.
instruction
jobs Package jobs is the session-scoped background-job registry behind the agent's background tools (bash run_in_background, task run_in_background) and the bash_output / kill_shell / wait tools.	Package jobs is the session-scoped background-job registry behind the agent's background tools (bash run_in_background, task run_in_background) and the bash_output / kill_shell / wait tools.
lsp Package lsp is a minimal Language Server Protocol client: it spawns a language server per language on demand, syncs queried documents from disk, and adapts a few read-only capabilities (definition, references, hover, diagnostics) to the tool.Tool interface.	Package lsp is a minimal Language Server Protocol client: it spawns a language server per language on demand, syncs queried documents from disk, and adapts a few read-only capabilities (definition, references, hover, diagnostics) to the tool.Tool interface.
mcpdiag
memory Package memory implements fairpeer's persistent memory.	Package memory implements fairpeer's persistent memory.
netclient Package netclient builds HTTP clients and proxy resolvers that share fairpeer's user-facing proxy settings.	Package netclient builds HTTP clients and proxy resolvers that share fairpeer's user-facing proxy settings.
nilutil
notify
outputstyle Package outputstyle adds a selectable "output style" — a block of persona / tone instructions appended to (or replacing) the system prompt — so the user can shift how the agent communicates without rewriting the system prompt.	Package outputstyle adds a selectable "output style" — a block of persona / tone instructions appended to (or replacing) the system prompt — so the user can shift how the agent communicates without rewriting the system prompt.
permission Package permission decides, per tool call, whether to allow it, deny it, or ask the user first.	Package permission decides, per tool call, whether to allow it, deny it, or ask the user first.
plugin Package-internal cache for MCP handshake results.	Package-internal cache for MCP handshake results.
ppttemplate Package ppttemplate provides PPT template rendering utilities.	Package ppttemplate provides PPT template rendering utilities.
proc
provider Package provider defines the model-backend abstraction and a registry mapping a provider "kind" to a factory.	Package provider defines the model-backend abstraction and a registry mapping a provider "kind" to a factory.
provider/anthropic Package anthropic implements the Anthropic Messages API provider (POST /v1/messages, SSE streaming) with a hand-written net/http client — no SDK.	Package anthropic implements the Anthropic Messages API provider (POST /v1/messages, SSE streaming) with a hand-written net/http client — no SDK.
provider/openai Package openai implements the OpenAI-compatible /chat/completions provider.	Package openai implements the OpenAI-compatible /chat/completions provider.
rag Package rag implements a knowledge-base store for the coWork profile: import documents, chunk them, and search them.	Package rag implements a knowledge-base store for the coWork profile: import documents, chunk them, and search them.
runtime
sandbox Package sandbox wraps a shell command in an OS-level jail so the model's `bash` calls are confined: it may read freely but write only inside the workspace (plus temp and toolchain caches) and reach the network only when allowed.	Package sandbox wraps a shell command in an OS-level jail so the model's `bash` calls are confined: it may read freely but write only inside the workspace (plus temp and toolchain caches) and reach the network only when allowed.
scheduler Package scheduler is a persistent, session-independent task scheduler for the coWork profile.	Package scheduler is a persistent, session-independent task scheduler for the coWork profile.
secret
serve Package serve exposes a control.Controller over HTTP: the typed event stream as Server-Sent Events, and the commands as small JSON POST endpoints.	Package serve exposes a control.Controller over HTTP: the typed event stream as Server-Sent Events, and the commands as small JSON POST endpoints.
skill Package skill loads invokable playbooks ("skills") from Markdown files.	Package skill loads invokable playbooks ("skills") from Markdown files.
sysproxy Package sysproxy resolves the OS-level proxy (Windows system/PAC settings) for a target URL.	Package sysproxy resolves the OS-level proxy (Windows system/PAC settings) for a target URL.
tool Package tool defines the Tool abstraction and a Registry.	Package tool defines the Tool abstraction and a Registry.
tool/builtin Package builtin provides fairpeer's compile-time built-in tools.	Package builtin provides fairpeer's compile-time built-in tools.
validation Package validation provides pre-write syntax validation for the editor tools (write_file/edit_file/multi_edit).	Package validation provides pre-write syntax validation for the editor tools (write_file/edit_file/multi_edit).