Skip to main content

Daita data operations agent

Daita

An autonomous data operations agent that learns your data systems.

Data systems are more than tables. Daita maps their structure, learns the business meaning you approve, and turns individual investigations into durable jobs and recurring checks. It answers from current evidence and carries out admitted actions. The goal is fully autonomous data operations within the access, budgets, and actions you authorize.

What Daita can do

  • Understand your data. Explore SQLite and PostgreSQL catalogs, follow relationships between resources, and run durable data profiles without changing source data.
  • Answer across systems. Query admitted databases, local CSV, TSV, JSON, NDJSON, and Parquet files, and remote MCP tools using current source evidence.
  • Carry work forward. Run durable task graphs or set assignments to run once or regularly. Review their results in the inbox while a Daita host is open.
  • Improve with use. Keep conversations, approved business knowledge, meanings tied to specific sources, and reusable Markdown procedures. An optional, explicit review can propose lessons from completed runs for you to accept.
  • Deliver useful output. Create reports backed by evidence and CSV or XLSX exports, including exact relational snapshots.
  • Take controlled action. Preview and approve scoped PostgreSQL updates or upserts, and admit specific remote MCP actions. Durable receipts support inspection and recovery when an outcome is uncertain.

Daita works with OpenAI, Anthropic, Gemini, Grok, Ollama, endpoints compatible with OpenAI's API, and supported model subscriptions.

Quick start

On supported macOS and Linux systems, install Daita with:

curl -fsSL https://daita-tech.io/install.sh | bash

The installer supplies its own Python runtime and starts onboarding when run in a terminal. Run daita in a new terminal to return later.

The first launch guides you through creating an agent, configuring a model, and optionally attaching a data source for reading. The launch directory becomes the default local working directory; foreground Files tools also accept explicit local paths.

Try asking:

How do our customers, orders, and payments tables relate?
Profile the orders and payments tables without changing them.
Compare paid revenue by region across our admitted sources.
Every Monday, summarize last week's revenue and put the result in my inbox.

Use /learn <material> to teach Daita a business definition or procedure with approval. /jobs shows durable work, /routines manages saved assignments, and /inbox shows their results. /artifacts previews, saves, and deletes stored files. /sources explores connections and relationships; @ narrows a question to one source. Type /help for the full command list.

Scheduled work progresses only while the terminal or daita host --agent <name> keeps that agent open. One host can open an agent home at a time.

Authority stays explicit

Database connections start with read access only. Daita validates SQL against the current catalog, checks tool calls against current scope and permissions, and treats source content, remote output, memory, and skills as untrusted input. Saved knowledge helps interpret data; it cannot grant access or override current source facts.

External effects require exact permissions and approval or a bounded standing grant. Native writes use a preview from the current run and transactional checks. Daita records effect receipts and never silently retries an uncertain action.

See relational writes, remote MCP connectivity, and action receipts and recovery for the exact limits.

Guides

Topic Guide
Local file access and edits Local computer files
Reports, exports, and provenance Artifacts
Source scope and retained context Context and scope
Scheduled work, outcomes, and hosting Scheduled routines
Durable investigations and task graphs Durable jobs
Learning, business meaning, and procedures Learning and semantics
Model setup and subscriptions Model sources
Python API walkthroughs Offline examples
Architecture and development Repository guide · Contributing
Private vulnerability reporting Security policy

Upgrade or uninstall

Close any running Daita terminal or host first. Run the Quick start command again to upgrade a managed installation. To remove it, run:

curl -fsSL https://daita-tech.io/install.sh | bash -s -- --uninstall

Agent state under ~/.daita is separate from the installation. Existing pipx installations still use pipx for upgrades and removal. Daita 0.19.0 and earlier belong to a different framework family and cannot be migrated into 1.x; preserve that state before upgrading. See local state compatibility.

License

MIT

Metadata

Release files for daita-agents 1.2.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for daita-agents 1.2.2
File Interpreter ABI Platform Actions
daita_agents-1.2.2-py3-none-any.whl Python 3 none any Details for daita_agents-1.2.2-py3-none-any.whl

Release files / daita_agents-1.2.2-py3-none-any.whl

Download URL daita_agents-1.2.2-py3-none-any.whl
Size 1.2 MB
Tags Python 3
SHA-256 checksum
How to use checksums
bbf966bcd19874333356dc84f859ea4337f8bfd181ed792621caa71c5f0099f0
BLAKE2b-256 checksum
How to use checksums
234d947695c79971bd9dcd40d808d241a97290635dab6afa77bc49cc091ec481
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.12

Security for daita-agents 1.2.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

1.2.2 This release

1 release file No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page