# Shiba > AI employees for the work behind your company. Describe a task, review the proposed action, approve repository execution, and inspect the files and changes returned. - Audience: solo builders and teams working across engineering, design, marketing, sales, operations, finance, research, and revenue planning. - Example outputs: a code change, design specification, campaign draft, account brief, operating checklist, pipeline summary, or draft budget using supplied figures. - Roles: Engineering, Marketing, Research, and Operations have built-in templates in this branch. Other responsibilities use custom employee instructions. Roles do not grant connected service access. - Execution: a GitHub repository is required even for non-code drafts. Human approval covers the exact requested work; external publishing is a separate supported/configured action. - Finance and revenue: draft analysis from supplied information, not bookkeeping sync, payroll, payments, autonomous CRM updates, or guaranteed financial outcomes. - Ownership and costs: free, open-source software (AGPL-3.0-only). Hosting and model usage are paid to the configured providers. No free managed plan is offered. - Status: under active development. Supported connections still need validation in the installation; see https://tryshiba.dev/docs/readiness/ for current verification limits. - Maintainer: Prince Pal (https://github.com/princepal9120) ## Links - Website: https://tryshiba.dev/ - Docs: https://tryshiba.dev/docs/overview/ - GitHub source (AGPL-3.0-only): https://github.com/princepal9120/shiba - Open-source Devin alternative guide: https://tryshiba.dev/open-source-devin-alternative/ - Why I am building Shiba: https://tryshiba.dev/why-shiba/ - Integrations (availability and configuration): https://tryshiba.dev/integrations/ - How work moves through Shiba: https://tryshiba.dev/network/ - Journal (building notes and guides): https://tryshiba.dev/blog/ - Join waitlist & contribute: https://tryshiba.dev/waitlist/ - Privacy notice: https://tryshiba.dev/privacy/ - Terms of use: https://tryshiba.dev/terms/ ## Installation and security - Edge Orchestrator: Cloudflare Workers + Agents SDK (TypeScript) routing requests at the edge. - Sandbox containers: Cloudflare Sandbox SDK + Containers provide isolated Linux containers per task. - Credentials: Supported provider-backed execution uses Worker-side egress authorization. Review the authentication-mode limits in https://tryshiba.dev/docs/configuration/ and https://tryshiba.dev/docs/security/ before choosing a connection. - State & Memory: Cloudflare Durable Objects + SQLite for human approval queues, live logs, and Vectorize for long-term memory. - Access: Designed to sit behind operator-configured Cloudflare Access for dashboard and runner endpoints. ## Documentation - [Claude Code](https://tryshiba.dev/docs/agents/claude-code/): Run Anthropic's Claude Code agent CLI in isolated Cloudflare Sandboxes with human approval gates. - [Codex](https://tryshiba.dev/docs/agents/codex/): Run OpenAI's Codex CLI autonomously within approval-gated Cloudflare Sandboxes. - [OpenCode](https://tryshiba.dev/docs/agents/opencode/): The default open-source coding agent harness in Shiba, supporting multi-provider AI Gateway routing. - [API reference](https://tryshiba.dev/docs/api/): Core dashboard routes, run requests, email operations, and approval transport for your Shiba installation. - [Approvals](https://tryshiba.dev/docs/approval-gates/): Check a proposed task before it starts, then approve it or ask for a change. - [Architecture](https://tryshiba.dev/docs/architecture/): Components, data flow, and unfinished integration boundaries. - [Automation configuration reference](https://tryshiba.dev/docs/automation-reference/): Trigger configuration, limits, and operator controls for Shiba automations. - [Automations](https://tryshiba.dev/docs/automations/): Set up recurring work and tasks that respond to events. - [Coding Harnesses](https://tryshiba.dev/docs/claude-code/): Coding harnesses, Pi and Grok model access, and current implementation and verification status. - [Configuration](https://tryshiba.dev/docs/configuration/): Current runtime defaults, optional secrets, and what harness verification means. - [Contributing](https://tryshiba.dev/docs/contributing/): Local verification, documentation contracts, and contribution boundaries. - [Costs](https://tryshiba.dev/docs/costs/): Understand what you pay for when you use or host Shiba. - [Threads and results](https://tryshiba.dev/docs/dashboard/): Give Shiba a task, approve the request, and review the result. - [Deployment preparation](https://tryshiba.dev/docs/deployment/): Prepare an account-owned installation without confusing local validation with live acceptance. - [Discord](https://tryshiba.dev/docs/discord/): Queue a Shiba task from Discord and review it before work starts. - [Discord setup](https://tryshiba.dev/docs/discord-setup/): Connect Discord interactions, register the task command, and configure approvers. - [Employees](https://tryshiba.dev/docs/employees/): Create an employee for work you repeat, then start a thread with its instructions. - [Your first task](https://tryshiba.dev/docs/getting-started/): Give Shiba a clear brief, approve the work, and review the result. - [GitHub](https://tryshiba.dev/docs/github/): Choose a repository, review changes, and decide whether to open a pull request. - [GitHub setup](https://tryshiba.dev/docs/github-setup/): Configure repository access, pull request publishing, and verified GitHub events. - [Grok Bot bridge](https://tryshiba.dev/docs/grokbot-bridge/): Pair Grok Bot SaaS connectors with Shiba founder skills — MCP task queue, status polling, and Slack digests on cron. - [Shiba vs hosted coding agents](https://tryshiba.dev/docs/hosted-alternatives/): Neutral comparison between self-hosting Shiba on your Cloudflare account and using a vendor-hosted coding-agent product — no lock-in claims, public facts only. - [Implementation status reference](https://tryshiba.dev/docs/implementation-status/): The dated engineering inventory and live-verification gaps for developers and installation owners. - [Launch](https://tryshiba.dev/docs/launch/): One-command deploy — pnpm bootstrap collects config, connects Cloudflare, builds, deploys, and prints the dashboard-only follow-ups. - [Linear](https://tryshiba.dev/docs/linear-agent-session/): Turn a delegated Linear issue into a task you can review in Shiba. - [Linear setup](https://tryshiba.dev/docs/linear-setup/): Configure Linear agent-session delivery and understand the current connection limits. - [Local development](https://tryshiba.dev/docs/local-development/): Run the dashboard, Worker, docs, and tests without overstating local or cloud evidence. - [Use from Claude Code](https://tryshiba.dev/docs/mcp/): Connect Claude Code (or any MCP client) to the Worker's /mcp gateway with a scoped bearer token and queue approval-gated coding runs. - [Model authentication modes](https://tryshiba.dev/docs/model-auth/): API keys, ChatGPT SIWC plan usage, and Claude compliance paths. - [Onboarding and setup](https://tryshiba.dev/docs/onboarding/): Prepare a local checkout and understand what has—and has not—been verified for live use. - [Welcome to Shiba](https://tryshiba.dev/docs/overview/): Get help with your work, from the first brief to the finished draft. - [Product status](https://tryshiba.dev/docs/product-status/): What you can try today and what to check before relying on it. - [Readiness and acceptance checklist](https://tryshiba.dev/docs/readiness/): Known gaps and evidence required before production use. - [Remote access](https://tryshiba.dev/docs/remote-access/): Connect your computer and choose it for approved local tasks. - [Remote access setup reference](https://tryshiba.dev/docs/remote-access-setup/): Operator configuration, command-line pairing, and local desktop troubleshooting. - [Pull request review](https://tryshiba.dev/docs/review/): Open a pull request, read the findings, and decide what to do with the changes. - [Review integration reference](https://tryshiba.dev/docs/review-reference/): Review transport, retries, and publishing boundaries for developers integrating with Shiba. - [Runtime costs and limits](https://tryshiba.dev/docs/runtime-limits/): Hosting references and configured runtime limits for installation owners. - [Choosing a sandbox provider (cost)](https://tryshiba.dev/docs/sandbox-cheapest-provider/): Neutral cost-oriented guide for picking Cloudflare, docker-local, or ComputeSDK backends (prices as of 2026-10-19). - [Write your own sandbox provider](https://tryshiba.dev/docs/sandbox-custom-provider/): Register a custom @shiba/sandbox adapter without forking shiba. - [Sandbox providers](https://tryshiba.dev/docs/sandbox-providers/): Choose where shiba runs agent containers — Cloudflare by default, or alternate backends for self-hosting. - [Sandbox SDK 1.0 migration](https://tryshiba.dev/docs/sandbox-sdk-1/): Phase 2a — move shiba from @cloudflare/sandbox 0.12 to 1.0 with outbound handlers and a custom Sandbox Durable Object. - [TypeScript SDK](https://tryshiba.dev/docs/sdk/): Use @shiba/sdk to call the public /api/v1 API from Node, Bun, Deno, browsers, and Workers. - [Security](https://tryshiba.dev/docs/security/): Authentication, credential boundaries, and known limitations. - [Self-hosting on Cloudflare](https://tryshiba.dev/docs/self-hosting/): Deploy Shiba into your own Cloudflare account with bring-your-own model keys, one deploy tool per environment, and no secret drift between Alchemy and wrangler. - [Skills](https://tryshiba.dev/docs/skills/): Save instructions you use often and add them to individual tasks. - [Slack](https://tryshiba.dev/docs/slack/): Ask Shiba for a task and review its approval request from Slack. - [Slack setup](https://tryshiba.dev/docs/slack-setup/): Install the Shiba Slack app, choose approvers, and connect channels to repositories. - [Swarm desk](https://tryshiba.dev/docs/swarm-desk/): Four lanes over the runs you already have, plus share links and Linear or Sentry paste. - [Task settings reference](https://tryshiba.dev/docs/task-settings/): Technical details for reasoning effort, repository setup, and per-task environment variables. - [Telegram](https://tryshiba.dev/docs/telegram/): Send a task to Shiba and approve it in your Telegram chat. - [Telegram setup](https://tryshiba.dev/docs/telegram-setup/): Connect a Telegram bot and choose who may approve work in each chat. - [Agent templates](https://tryshiba.dev/docs/templates/): Browse, add, copy, and import Shiba agent template packs. - [HTTP Triggers (iPhone Shortcuts)](https://tryshiba.dev/docs/triggers/): Queue approval-gated coding tasks from an iPhone Apple Shortcut or any HTTP client via POST /api/trigger. - [Common questions](https://tryshiba.dev/docs/troubleshooting/): Find the next step when you're stuck with a task or your workspace. - [Waitlist and contributor signups](https://tryshiba.dev/docs/waitlist/): How the public Shiba signup form stores entries and how the operator exports or removes them. ## Journal Building notes and guides. Dated articles describe the product and evidence available when they were written. - [An open-source coding agent alternative: a guide to Shiba](https://tryshiba.dev/blog/open-source-coding-agent-alternative/) (Guide): A grounded look at Shiba as an open-source alternative to hosted coding agents, covering what is inspectable, what is gated, and what is not yet proven. - [Why self-host a coding agent at all](https://tryshiba.dev/blog/why-self-host-a-coding-agent/) (Note): Self-hosting a coding agent only matters if the harness, the model, and the account are all yours, because that combination is what the seat fee never buys. - [Approvals in Slack, not another dashboard](https://tryshiba.dev/blog/approvals-in-slack-not-another-dashboard/) (Guide): A mention in a Slack thread becomes a pending approval card, and a click from a named approver on that card is what allows a run to start. - [What a failed run looks like](https://tryshiba.dev/blog/what-a-failed-run-looks-like/) (Journal): Shiba reports failures with a real exit code, a bounded stderr tail, and an error envelope, and never dresses a failure up as a success. - [Harness and model choices, and what they do not tell you](https://tryshiba.dev/blog/harness-and-model-choices/) (Guide): Shiba lets each run pick a coding harness and model, refuses bad pairings before execution, and publishes no cost or latency numbers. - [Credentials that never enter the container](https://tryshiba.dev/blog/credentials-that-never-enter-the-container/) (Guide): A walk through what Shiba hands to the sandbox, what it holds back, and where each key is actually attached. - [Why the approval gate comes before the container](https://tryshiba.dev/blog/approval-gate/) (Guide): Shiba stops a planned coding run at a human decision point, and that gate is the safety invariant the rest of the system is built around.