# Tale — Open-source project workspace where teams assign tasks to AI coding agents running in sandboxes.

> Source: The Agents Index — https://theagentsindex.com/tale (structured, researched, re-verified)
> Facts last verified: 2026-10-03

Tale is an MIT-licensed project workspace in which teams plan tasks and delegate them to AI agents that run on coding harnesses inside sandboxes. It is sold as a self-hosted Community edition or a professionally operated Enterprise deployment.

| Fact | Value |
| --- | --- |
| Website | https://tale.dev |
| Price | from CHF 12/month billed monthly, CHF 10/month billed annually |
| Free tier | Yes |

Community edition: free to self-host under MIT with every product feature, but self-hosted only and supported solely on GitHub; cloud hosting, ISO 27001-certified operation, PII redaction, custom DPA and email/phone/remote support are Enterprise-only, and model inference is billed by your own AI provider.

## What it does

**Writes code** · Yes — Project agents run on a coding harness that reads and writes files and runs commands in a sandbox, so new code is authored from the task brief.

> A harness is the coding program that runs an agent's session in a sandbox. It asks the model what to do, reads and writes files, runs commands, and returns a report.
> — https://docs.tale.dev/platform/agents/harnesses · Opening paragraph, 'Choose an agent runtime'

**Reviews code** · Partly — A separate agent can be granted a review tool to judge another agent's completed result and its evidence, but the vendor scopes this to task deliverables and says Tale does not verify GitHub state itself.

> If GitHub work is involved, the reviewer must check the referenced commit and checks itself: Tale records its evidence but does not independently verify GitHub's current state or merge a pull request as part of the verdict.
> — https://docs.tale.dev/platform/projects/task-automation · Section 'Set up an independent reviewer', final paragraph

**Whole-codebase work** · Partly — Each project agent keeps a persistent sandbox workspace and may clone repositories, but the vendor documents no repository-wide indexing or migration capability and caps runs by time and sandbox capacity.

> A project agent reuses a persistent workspace across its tasks. Task attachments are available read-only under `/agent/inputs/<task>/attachments/`
> — https://docs.tale.dev/platform/agents/harnesses · Section 'Understand files and connected tools', first paragraph

**Ships it** · No — Work stops at collected deliverables and a review verdict; the vendor states plainly that a review does not merge a pull request, and connector writes are refused to agents.

> Tale records its evidence but does not independently verify GitHub's current state or merge a pull request as part of the verdict.
> — https://docs.tale.dev/platform/projects/task-automation · Section 'Set up an independent reviewer', final paragraph

**Done-for-you service** · Partly — You run the software, but the Enterprise plan adds professional operation, installation and maintenance, with consulting and custom development charged hourly rather than a built-for-you agent.

> Everything in Community, professionally run — installation, maintenance, and support, in the cloud or on your hardware.
> — https://tale.dev/pricing · Enterprise plan card, description line

_Not established: Debugs and fixes, Writes tests_

## How much it does unattended

**Runs autonomously** · Partly — An agent completes a task unattended and posts its report, but it must stop at In review and cannot approve its own result, so a person or a separate reviewer agent closes the loop.

> Move the task to **Done** only when you accept the work. Tale records the decision; the implementation agent cannot approve its own result.
> — https://docs.tale.dev/platform/projects/task-automation · Section 'Read and accept the result'

**Multi-agent** · Yes — A manager agent granted the delegation tool starts other project agents on ready work, and a second agent can review the first's result within delegation and capacity limits.

> An equipped manager agent can start other agents on ready tasks in the same project, within delegation, access and capacity limits.
> — https://tale.dev/platform/agents · 'What can agents do?' list, item 03 'Delegation between agents'

**Agent permissions** · Partly — Each agent is granted explicit tools, skills and secrets, connector writes are refused to the agent broker, and external automation writes pause for approval by default.

> Automation connector writes pause when your policy requires approval. External writes require it by default; reads and internal platform writes follow different rules.
> — https://tale.dev/platform/governance · 'What does governance cover?' list, item 01 'Approve connector writes'

## Which models it runs on

**Claude** · Yes — Claude Code is a supported harness with managed or direct credentials, and Anthropic subscriptions are served through a subscription broker as well as API keys.

> Subscription brokers support Anthropic subscriptions through Claude Code and OpenAI ChatGPT subscriptions through Codex.
> — https://docs.tale.dev/platform/admin/providers · Section 'Connect a subscription broker', opening sentence

**GPT** · Yes — Codex runs as a supported harness and OpenAI is listed among the cloud providers Tale serves, with ChatGPT subscriptions usable through the broker.

> We support OpenAI, Anthropic, Gemini, and other major cloud AI providers in CH & EU — no markup; usage is billed at the provider's rates.
> — https://tale.dev/pricing · Compare Plans table, 'Cloud AI providers' row

**Open models** · Yes — A custom provider credential points at a self-hosted model server such as vLLM or Ollama, or any internal gateway speaking the OpenAI or Anthropic API.

> An organization can connect an endpoint the shipped catalog does not list: a self-hosted model server such as vLLM or Ollama, or an internal gateway that speaks the OpenAI or Anthropic API.
> — https://docs.tale.dev/platform/admin/providers · Section 'Define a custom provider', opening sentence

**Model choice** · Yes — The buyer picks the provider credential and pins a model per agent; the vendor ships no model of its own and admins restrict access per person or team.

> Search **Model** by name or API ID. The same model can appear once per provider; read the provider on the entry before selecting it. Selecting an entry pins that pair for future runs.
> — https://docs.tale.dev/platform/projects/project-agents · Configure the agent, step 'Choose the model and provider'

## Where you use it

**In your editor** · Partly — There is no Tale IDE extension; instead opencode and Claude Code are pointed at Tale as a model provider and MCP tool server from your own editor.

> The model endpoints let opencode, Claude Code, and scripts built on the OpenAI or Anthropic SDKs use your organization's approved models as their model provider.
> — https://docs.tale.dev/develop/use-tale-from-your-editor · Section 'Use your organization's models from your tools', opening paragraph

**On the command line** · Partly — A Tale CLI exists for deployment and configuration projects, and terminal agents can call the REST and MCP endpoints, but agent runs themselves are driven from the web app.

> `tale update` updates the CLI within its current release line and refreshes generated project content.
> — https://docs.tale.dev/develop/ai-assisted-development · Section 'Keep the mirror fresh', first sentence

**In your pipeline** · Partly — Automations start from webhooks and schedules and a GitHub connector reads and manages pull requests, but agents get read-only connector actions and nothing runs as a CI job.

> GitHub | Read or manage repositories, issues and pull requests. | Token.
> — https://docs.tale.dev/platform/connectors/overview · 'Choose the connection for the task' table, GitHub row

**In a browser** · Yes — Tale is a hosted or self-hosted web workspace: boards, agent tabs, sandbox file and live-browser panes all render in the application.

> Agents that run on a harness work in a sandbox — browse the workspace tree, open the file they just wrote, and watch the live browser as they click through the result.
> — https://tale.dev/platform/agents · Section '04 Sandbox — Watch the agent in Files and Live'

## Whose machine it runs on

**Self-hosted** · Yes — Self-hosting is the Community edition's only deployment mode and remains an option on Enterprise, on your own hardware or in Tale's cloud.

> Every product feature, free to self-host — support happens on GitHub.
> — https://tale.dev/pricing · Community plan card, description line

**Open source** · Yes — The platform is published under the MIT licence with its repository on GitHub, and the vendor states all product features are included in Community.

> Yes. The Community edition is free to self-host under the MIT license.
> — https://tale.dev/pricing · Pricing FAQ, answer to 'Is Community free?'

## What it costs to run

**How it meters** · Yes — Enterprise is priced per user per month plus storage per terabyte, with model inference billed separately by whichever AI provider you connect.

> All product features are included in both Community and Enterprise. Additional cloud AI model usage is billed at your provider's rates.
> — https://tale.dev/pricing · Note under the Enterprise plan card

**Free tier** · Yes — The Community edition is a standing free self-hosted option with every product feature, needing no sales call; support happens on GitHub.

> The MIT-licensed Community edition is free to self-host.
> — https://tale.dev/pricing · Page introduction, second sentence

**API access** · Yes — A documented REST API with an OpenAPI schema covers projects, tasks, agents, runs and threads using a personal API key and organization header.

> The REST API lets you read and change Tale resources with an API key: projects, files, tasks, automations, runs and chat threads.
> — https://docs.tale.dev/develop/api-reference · Opening paragraph, 'API reference'

**MCP server** · Yes — Tale serves its own MCP endpoint over HTTPS JSON-RPC so an external agent can discover tools, retrieve knowledge and run automations.

> Connect an MCP client when an agent needs to discover Tale's tools, retrieve knowledge, or build and run automations. The connection uses the same API key and organization scope as REST.
> — https://docs.tale.dev/develop/mcp-endpoint · Opening paragraph, 'MCP endpoint'

**Bring your own key** · Yes — Organisations supply their own provider API keys, environment-variable secrets, vendor subscriptions or a broker, and can run the whole platform on their own infrastructure.

> **API key** | The provider's secret key | Standard metered API access. The stored secret is encrypted and later shown only as a masked fragment.
> — https://docs.tale.dev/platform/admin/providers · 'Choose an authentication method' table, API key row

## Buying it for a team

**A company can buy it** · Yes — Enterprise is sold to organisations at a published per-user monthly rate with a service agreement, custom DPA and professional operation, bought through the contact route.

> Tale Enterprise is CHF 12 (EUR 14) per user/month, with two months free on yearly billing. Storage is CHF 10 (EUR 12) per TB/month.
> — https://tale.dev/pricing · Pricing FAQ, answer to 'How much does Tale Enterprise cost?'

**Seat model** · Yes — Billing is strictly per user per month; the calculator's slider runs from a single user upward and the vendor publishes no seat minimum or ceiling.

> 25 users × CHF 12/user/month
> — https://tale.dev/pricing · Enterprise plan card, price breakdown line beneath the calculator

**Pooled budget** · Yes — Budget rules can be scoped to the whole organization, a team, a role, a user or one API key, so spend is pooled or individual as configured.

> Use a role for a group such as Editors, a team for a shared workload, a user for an individual, an API key for one credential, or the organization for a shared ceiling.
> — https://docs.tale.dev/platform/admin/governance/policies-and-limits · Section 'Add a spending budget', step 2

**Admin controls** · Yes — Admins set organization rules, tool grants, model access, content filters and spending caps, and can disable task automation for the whole organization.

> Configure shared instructions, access policies, and agent tool grants. Enforced permissions and filters complement the instructions you give a model.
> — https://tale.dev/platform/governance · 'What does governance cover?' list, item 04 'Set organization rules'

**Audit log** · Yes — An audit trail records actions and approval decisions with their actor and time, and the vendor documents audit-log integrity for tamper evidence.

> Inspect recorded actions and approval decisions with their actor and time. Use the audit log to investigate what happened.
> — https://tale.dev/platform/governance · 'What does governance cover?' list, item 03 'Read the audit trail'

**Single sign-on** · Yes — SAML 2.0, generic OIDC, OAuth2 and Microsoft Entra ID are all supported, with SCIM provisioning and group-to-team sync alongside sign-in.

> Enterprise SSO lets members sign in through your identity provider (IdP). SCIM lets that provider create, update, and deactivate members without waiting for them to sign in.
> — https://docs.tale.dev/platform/admin/enterprise-sso · Opening paragraph, 'Enterprise SSO and provisioning'

## What happens to your code

**Opt out of training** · Yes — The DPA commits contractually to no training on customer payloads, binds AI sub-processors to the same, and says a product toggle cannot vary it.

> Tale does not use Personal Data — including prompts, inputs, outputs, embeddings, audio payloads, images, or any derived artifacts — to train, fine-tune, evaluate, benchmark, or otherwise improve any AI or machine-learning model
> — https://tale.dev/legal/data-processing-agreement · Section 5.1 'No training, fine-tuning, or model improvement'

**Data residency** · Yes — Cloud hosting sits in Switzerland for Swiss customers and the EU for everyone else, with AI inference kept inside the EU/EEA; self-hosters choose their own locations.

> Tale hosts the platform and stores Personal Data in Switzerland for Swiss customers and in the European Union for all other customers.
> — https://tale.dev/legal/data-processing-agreement · Section 11.1 'Processing locations'

**Getting out** · Yes — On termination the customer can demand return of all data in a machine-readable format or certified deletion, and there are no credits to expire.

> returns all Personal Data to the Customer in a commonly used, machine-readable format; or b) securely deletes all Personal Data and provides written confirmation of deletion.
> — https://tale.dev/legal/data-processing-agreement · Section 13.2 'On termination'

**Certifications** · Partly — Ruler GmbH holds ISO/IEC 27001 covering Tale Enterprise and offers a GDPR/FADP DPA with a custom DPA on Enterprise, but no SOC 2 report is published.

> Ruler GmbH, the company behind Tale, is ISO/IEC 27001 certified. The certification covers its Tale Enterprise offering and services.
> — https://docs.tale.dev/cloud/trust-and-compliance · Opening paragraph, 'Security and compliance'

## Provenance

Every fact above comes from the vendor. Nothing here is independently corroborated yet.

## Coverage elsewhere

None recorded. Every fact above comes from the vendor and is not independently corroborated yet.

## What changed

- 2026-10-03 — Capabilities · Commercial terms · Free tier · Price

## Correcting this record

Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Tale, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth.

Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.