sprout

command module

v0.32.2 Latest Latest Go to latest Published: Oct 18, 2026 License: Apache-2.0 Imports: 1 Imported by: 0

Details

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

Repository

github.com/sprout-foundry/sprout

Links

Open Source Insights

README ¶

Sprout

An open-source AI coding agent for your terminal and browser. Point it at a project, describe what you want, and review what it changed — every file edit is tracked and can be rolled back, and risky commands ask before they run.

Install

macOS / Linux:

curl -fsSL https://raw.githubusercontent.com/sprout-foundry/sprout/main/scripts/install.sh | sh

Windows (PowerShell 5.1+):

irm https://raw.githubusercontent.com/sprout-foundry/sprout/main/scripts/install.ps1 | iex

The installer verifies the release checksum. Set SPROUT_INSTALL_DIR to choose where it goes — see scripts/install.sh. Run it again to upgrade.

Get it working (about 5 minutes)

Sprout runs on an AI model from a provider you choose. You pay the provider for what you use — usually cents per task.

1. Get an API key from one of these:

DeepInfra — sign up, add a little credit, create an API key.
OpenRouter — the alternative: sign up, add credit, create an API key. One key gives you hundreds of models.

2. Open your project and start Sprout:

cd path/to/your/project
sprout

The first run asks which provider you use, takes your key and checks it. You can also set the key ahead of time with sprout keys set deepinfra (or openrouter), or by exporting DEEPINFRA_API_KEY / OPENROUTER_API_KEY.

3. Pick a model. We suggest DeepSeek V4.1 Flash or GLM-5.3-Flash — both fast, strong at coding, and inexpensive. In Sprout, type one of:

/model deepseek-ai/DeepSeek-V4.1-Flash   # on DeepInfra
/model zai-org/GLM-5.3-Flash             # on DeepInfra
/model deepseek/deepseek-v4.1-flash      # on OpenRouter
/model z-ai/glm-5.3-flash                # on OpenRouter

/model on its own lists every model your provider offers; /provider select switches providers.

4. Ask for something. For example:

Explain how this project is organized
Add a test for <some function>
Find and fix the failing test

Type /help to see every command.

Using Sprout

In the terminal. sprout starts an interactive session in the current directory. It creates a .sprout/ folder in your project for its project settings.

In the browser. While sprout is running it also serves a web UI — chat, editor, file tree, terminal and git — at the address it prints (http://localhost:56000 by default). To run only the web UI, use sprout agent -d. To keep it running in the background on macOS or Linux, install it as a service with sprout service install.

Other commands:

sprout plan                  # propose a plan first, build after you approve
sprout review                # AI review of your staged changes
sprout commit                # write a commit message for staged changes
sprout pr                    # open a pull request for the current branch
sprout search "rate limit"   # search your past sessions

Undo and safety

/changes lists the files Sprout changed in this session.
/rollback lists revisions; /rollback <revision-id> reverts one. From your shell, sprout log shows the same history.
Risky commands and edits ask for your approval before they run. How much asks is set by a risk profile:

Profile	Behavior
`readonly`	Read-only operations only; changes are blocked
`cautious`	Most operations ask first
`default`	Reads and common edits run; destructive operations ask
`permissive`	Almost everything runs without asking
`unrestricted`	Nothing asks; only critical patterns (`rm -rf /`, fork bombs) are blocked

sprout agent --risk-profile=cautious    # for one session; set "risk_profile" in your config to keep it

Sprout calls your provider's API, which costs money per token; you can see a session's cost in the web UI. Full security model: docs/SECURITY.md.

What's included

Agent and personas — coder, tester, reviewer, an orchestrator that runs subagents in parallel, and more; steer or cancel mid-task
Web UI — chat, editor, file tree, terminal, git, settings, cost meter; built into the binary
Providers — DeepInfra, OpenRouter, OpenAI, Z.AI, GLM Coding Plan, DeepSeek, Mistral, MiniMax, Cerebras, Chutes, Ollama Cloud, LM Studio, self-hosted Ollama, and any OpenAI-compatible endpoint (sprout custom add)
History — every file change tracked and reversible; full-text search across past sessions
Skills and MCP — instruction packs (sprout skill) and external tool servers (sprout mcp)
LSP — diagnostics, definitions and rename (sprout lsp)
Automations — JSON-defined workflows (sprout automate)

Configuration

Settings are layered — global defaults, optional per-project overrides, then the current session — and every setting is also editable in the web UI's settings panel. See docs/CONFIGURATION.md and docs/FALLBACKS.md.

Contributing

Requirements: Go 1.26+ and Node.js 22+.

git clone https://github.com/sprout-foundry/sprout.git
cd sprout
make prepare-grammars        # editor language grammars (Make targets run it automatically)
make build-all               # WASM shell + embedded web UI + Go binary + @sprout-foundry/workspace
./sprout                     # run from the repo root

On Windows without make/bash, powershell -File scripts\prepare-grammars.ps1 then go build . builds the binary; the agent's shell tool uses Git for Windows' bash when installed (cmd.exe otherwise).

Command	What it does
`make build-all`	Build everything (WASM, web UI, binary, workspace package)
`make test-smoke`	Fast smoke suite
`go test ./...`	Go unit tests
`make test-webui-vitest`	Web UI unit tests (vitest)
`npx playwright test --project=webui test/webui/<spec>.spec.ts`	Web UI e2e (backend + Vite stack auto-starts)
`make vet && make fmt-check && make lint`	Pre-PR gates

Web UI development with hot reload: run the backend (./sprout agent --daemon) next to the Vite dev server (cd webui && npm run dev) and edit webui/src/** with instant HMR at http://localhost:3000 — no UI rebuild needed. ./sprout serve swaps in a canned-response mock LLM so no provider credentials are required. Full instructions and gotchas: webui/DEVELOPMENT.md.

Agent guidance for working in this repo: AGENTS.md. Contribution rules: CONTRIBUTING.md. Authoritative specs: roadmap/ (SP-###.md).

Architecture in 30 seconds

CLI (sprout, Go) — cobra command tree (agent, plan, commit, review, pr, search, service, …)
Web UI (webui/, React 18 + Vite + TypeScript) — built by make deploy-ui, embedded in the binary
@sprout/ui (packages/ui) — shared component library, npm install @sprout/ui consumes it standalone; @sprout/events ships the event schemas
@sprout-foundry/workspace (packages/workspace) — the composition API a host mounts (SP-160); built by make build-workspace-package into packages/workspace/dist
Provider catalog (pkg/providercatalog/providers.json) — embedded in the binary, refreshed from GitHub at startup
Sister project sprout-foundry pins a SPROUT_VERSION and ships the binary in Docker images — integration contract: docs/FOUNDRY_CHAT_CONTRACT.md

Full layout and data flow: docs/ARCHITECTURE.md. Platform matrix (Linux/macOS/Windows/WASM/no-CGO): docs/PLATFORM_SUPPORT.md.

Documentation

Document	Description
CLI Reference	All commands, flags, slash commands, personas, tools
Onboarding	First-run setup in more detail
Configuration	Config files, environment variables, CI/CD
Fallbacks	Every default, inheritance, and fallback resolution chain
Architecture	Package layout, data flow, workspace files
Security	Risk profiles, tool call classification, security model
Personas	Persona system, risk model, custom personas
MCP Integration	MCP server setup, configuration, troubleshooting
Agent Workflow	Config-driven workflow sequences
Service / Daemon	Run sprout as a long-lived HTTP/WS service
Local LLM (MLX)	Run a model on your Mac (Apple Silicon), no API key
Provider Catalog	Provider catalog system and model metadata
Provider Registry	Remote provider registry, community provider PRs
LSP Architecture	Language server integration
Component Library	@sprout/ui usage and architecture
Testing	Test strategy, categories, commands
Changelog	Per-release commit log
Roadmap	Authoritative spec docs (`SP-###.md`) for planned and shipped work

License and support

Apache 2.0 — see LICENSE. Report issues at github.com/sprout-foundry/sprout/issues.

Documentation ¶

Rendered for

There is no documentation for this package.

Source Files ¶

View all Source files

main.go

Directories ¶

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

Path	Synopsis
cmd Agent command for sprout	Agent command for sprout
api_inventory command
audit_pricing command
benchmark-suite-check command Command benchmark-suite-check validates a benchmark task suite directory against the on-disk task fixture format, so a suite can be checked WITHOUT running the (priced) benchmark.	Command benchmark-suite-check validates a benchmark task suite directory against the on-disk task fixture format, so a suite can be checked WITHOUT running the (priced) benchmark.
enrich_registry command Command enrich_registry annotates freshly-generated canonical model files with capability-probe results.	Command enrich_registry annotates freshly-generated canonical model files with capability-probe results.
genapi command genapi writes the versioned OpenAPI contract (docs/api/openapi.yaml) from the Huma operations registered in pkg/webui.	genapi writes the versioned OpenAPI contract (docs/api/openapi.yaml) from the Huma operations registered in pkg/webui.
generate_studio_providers command Command generate_studio_providers derives the STUDIO_PROVIDERS snapshot embedded in the studio bridge (sprout-studio/shared/studio-bridge.js) from the provider registry, eliminating the hand-maintained copy that had drifted from the real provider configs.	Command generate_studio_providers derives the STUDIO_PROVIDERS snapshot embedded in the studio bridge (sprout-studio/shared/studio-bridge.js) from the provider registry, eliminating the hand-maintained copy that had drifted from the real provider configs.
model_probe command Command model_probe runs the capability probe against a single provider/model and prints the result as JSON.	Command model_probe runs the capability probe against a single provider/model and prints the result as JSON.
model_registry_server command Command model_registry_server runs a lightweight static file server for serving per-provider model JSON files.	Command model_registry_server runs a lightweight static file server for serving per-provider model JSON files.
refresh_provider_catalog command
sync_provider_configs command Command sync_provider_configs updates the embedded provider config models.available_models field to match the canonical registry.	Command sync_provider_configs updates the embedded provider config models.available_models field to match the canonical registry.
validate_registry command Command validate_registry checks every providers/*.json against the runtime schema before the publish workflow uploads to GitHub Pages.	Command validate_registry checks every providers/*.json against the runtime schema before the publish workflow uploads to GitHub Pages.
wasm command
internal
testgit Package testgit redirects git's config lookup inside test binaries so a subprocess git call can never read or write the developer's real ~/.gitconfig or /etc/gitconfig.	Package testgit redirects git's config lookup inside test binaries so a subprocess git call can never read or write the developer's real ~/.gitconfig or /etc/gitconfig.
pkg
agent Package agent: delegation getters and setters for sub-managers.	Package agent: delegation getters and setters for sub-managers.
agent/approvals Package approvals: the ApprovalAgent seam (SP-141 phase 3, increment 5).	Package approvals: the ApprovalAgent seam (SP-141 phase 3, increment 5).
agent/changes Auto-skip learning for ChangeTracker shell-mutation tracking.	Auto-skip learning for ChangeTracker shell-mutation tracking.
agent/subagents Package subagents: the process-wide active-subagent counter and the terminal display-prefix builder (SP-141 phase 4, increment 1).	Package subagents: the process-wide active-subagent counter and the terminal display-prefix builder (SP-141 phase 4, increment 1).
agent/workflow Package workflow provides the in-process workflow runner for TODO-loop workflows.	Package workflow provides the in-process workflow runner for TODO-loop workflows.
agent_api Package agent_api: LM Studio model-list discovery.	Package agent_api: LM Studio model-list discovery.
agent_audit Package agent_audit defines the facts-only per-call audit events the agent emits for every model call and tool execution, and the process-wide sink that records them.	Package agent_audit defines the facts-only per-call audit events the agent emits for every model call and tool execution, and the process-wide sink that records them.
agent_commands
agent_providers provider_config_timeouts.go — timeout + limit accessor methods on ProviderConfig, split from provider_config.go.	provider_config_timeouts.go — timeout + limit accessor methods on ProviderConfig, split from provider_config.go.
agent_tools design_export_paths.go — output-path confinement + artifact-relocation helpers for the design_export_tokens handler, split from design_export_handler.go.	design_export_paths.go — output-path confinement + artifact-relocation helpers for the design_export_tokens handler, split from design_export_handler.go.
agent_tools/computer_use Package computer_use denylist loader.	Package computer_use denylist loader.
apiconformance Package apiconformance is the conformance suite for the versioned backend contract in docs/api/openapi.yaml.	Package apiconformance is the conformance suite for the versioned backend contract in docs/api/openapi.yaml.
ast Package ast (continued) — body extraction.	Package ast (continued) — body extraction.
automate Package automate provides shared workflow discovery and validation for the automate/ feature used by both the CLI (cmd/automate.go) and the agent tool layer (pkg/agent/tool_handlers_automate.go).	Package automate provides shared workflow discovery and validation for the automate/ feature used by both the CLI (cmd/automate.go) and the agent tool layer (pkg/agent/tool_handlers_automate.go).
benchmark delegation.go — the benchmark's delegation axis: the run mode that selects how a run's work is delegated (coder only, worker delegation, or economy mode) and the per-run delegation metrics the report compares.	delegation.go — the benchmark's delegation axis: the run mode that selects how a run's work is delegated (coder only, worker delegation, or economy mode) and the per-run delegation metrics the report compares.
buildinfo Package buildinfo holds build-time metadata shared by the CLI, the WebUI bootstrap payload, and the update checker.	Package buildinfo holds build-time metadata shared by the CLI, the WebUI bootstrap payload, and the update checker.
clihooks Package clihooks holds tiny callback hooks that wire higher-level CLI behavior (e.g.	Package clihooks holds tiny callback hooks that wire higher-level CLI behavior (e.g.
cliui Package cliui hosts the CLI terminal display helpers extracted from cmd/.	Package cliui hosts the CLI terminal display helpers extracted from cmd/.
codecompletion
codegraph
codereview
configuration Package configuration: command categorization and force-flag detection.	Package configuration: command categorization and force-flag detection.
console Reasoning (collapsed "Thinking" header) methods for AssistantTurnRenderer, split from assistant_turn_renderer.go.	Reasoning (collapsed "Thinking" header) methods for AssistantTurnRenderer, split from assistant_turn_renderer.go.
credentials Encryption and decryption operations	Encryption and decryption operations