AI Tool Atlas

CodingDocumented overviewLast checked Sep 29, 2026

Runsight: Pricing, Features & Best Use Cases

Open-source, YAML-first workflow engine for AI agents with Git-native workflows, per-run cost tracking and built-in evals.

Runsight homepage screenshot, coding AI tool

Our take: Runsight suits developers who want agent workflows kept as YAML files in Git with per-run cost caps, run locally with their own API keys.

Main limitation: Requires CLI and YAML comfort and your own model API keys; no hosted version or pricing page was found.

By the AI Tool Atlas editors. Based on the vendor's official documentation; we have not tested this product hands-on. What “documented overview” means

Who Runsight fits

Runsight suits developers who want agent workflows kept as YAML files in Git with per-run cost caps, run locally with their own API keys. Based on the vendor's documentation, not on a hands-on test.

Best for: Developers

Also a good fit for

  • AI engineering teams

Consider another tool if requires technical setup (CLI, YAML)

Or if

  • Requires your own model API keys
  • No pricing page found (/pricing returns 404)

Runsight pricing and usage limits

Prices checked Sep 29, 2026 on the official US pricing page. Prices and limits change often.

Free access and trial conditions

Open source; started with one command, no signup or cloud account per the homepage.

Not applicable.

Paid plans: what changes as you upgrade

Open source (self-hosted) web

Free source (opens in a new tab)

One_time

Open source; no paid plans found (pricing page returns 404).

For: Developers

Limits

  • Runs on your machine
  • Your own API keys

Includes

  • YAML workflows
  • Cost tracking and budget caps
1 more
  • Built-in assertions

Usage caps, credits and extra costs

You pay your model providers for API usage.

Not applicable.

What is Runsight, and how does it work?

Runsight is an open-source workflow engine where AI agent workflows are YAML files on disk, versioned in Git and also viewable on a visual canvas. It is started with the uvx runsight command. It tracks cost per run, supports budget caps, assertions for evaluating outputs, and pausing or killing runs. It is self-hosted and uses your own API keys.

Runsight homepage — screenshot of the coding AI tool, Oct 2, 2026
Runsight's official site: how the product presents itself. runsight.ai, captured Oct 2, 2026

What you provide and what you get back

How Runsight works: what you provide, where it runs and what you get backYOU PROVIDEYour inputYAML workflow filesYour model API keysAgent definitions (souls) and too…CODINGRunsightmacos, linux, windowsfree tier, paid price not publish…free tier availablechecked Sep 29, 2026YOU GET BACKThe resultExecuted agent runs with cost and…Git-committed workflow filesAssertion resultsinput → Runsight → output · from the vendor's own pages, no hands-on test unless stated
What you give Runsight and what comes back, from its documentation

You provide

  • YAML workflow files
  • Your model API keys
  • Agent definitions (souls) and tools

You get back

  • Executed agent runs with cost and latency per block
  • Git-committed workflow files
  • Assertion results

Where you can use it

Platforms
macos, linux, windows, cli
Account
Not required
Install
uvx runsight (CLI)
Workflow format
YAML in Git
Block types
linear, gate, code, loop, workflow
License
Open source

Where Runsight is most useful

Scenarios are worked out by our editors from the documentation; the actual result depends on your input and plan.

Defining multi-step agent pipelines as code

Who
Developers building agent pipelines
You provide
A workflow.yaml with blocks and souls
Expected result
A runnable multi-step agent pipeline stored in Git
Plan needed
Open source
Limitation
Needs YAML and CLI skills source (opens in a new tab)

Capping spend on agent runs

Who
Teams worried about API bills
You provide
A budget block with max_cost and max_tokens
Expected result
Execution is killed before it exceeds the cap
Plan needed
Open source
Limitation
Caps are as accurate as provider pricing data source (opens in a new tab)

Regression-testing agent outputs

Who
AI engineers
You provide
Assertions on block outputs
Expected result
Regression checks across runs
Plan needed
Open source
Limitation
Eval design is left to the user source (opens in a new tab)

What Runsight does, and where it stops

Runsight homepage — screenshot of the coding AI tool, Oct 2, 2026
Feature overview on the official site. runsight.ai, captured Oct 2, 2026

YAML workflow definitions

Defines agent pipelines as YAML files with linear, gate, code, loop and workflow blocks. Lets teams review agent changes like code.

Access: Open sourceLimit: Requires learning the schema source (opens in a new tab)

Canvas and YAML views in sync

Keeps a visual canvas and the Monaco YAML editor showing the same workflow state. Lets you edit either way without drift.

Access: Local web UILimit: Runs on localhost source (opens in a new tab)

Per-run cost tracking and budget caps

Tracks cost per block and per run and can stop execution at a hard cap. Prevents surprise bills.

Access: Open sourceLimit: Depends on model pricing data source (opens in a new tab)

Built-in assertions and eval framework

Runs assertions and transform hooks on block outputs for regression testing. Replaces "looks good" with checks.

Access: Open sourceLimit: No built-in test datasets mentioned source (opens in a new tab)

Pros, cons and deal-breakers

Reasons to choose Runsight

Check before paying for Runsight

How to use Runsight for building a cost-capped agent workflow in YAML

Runsight documentation — screenshot of the coding AI tool, Oct 2, 2026
Runsight documentation: where the steps below come from. runsight.ai, captured Oct 2, 2026

Set up Runsight

  1. Install uv and run uvx runsight in a project folder
  2. Open localhost:8000
  3. Add your model API key in the onboarding flow
  4. Create a workflow YAML file or drag blocks on the canvas

Example: building a cost-capped agent workflow in YAML

  1. Define research and review blocks with soul references
  2. Add a budget with max_cost
  3. Commit the YAML to Git
  4. Run runsight run research-pipeline

Review and export the result

  1. Check the per-block cost matches the cap
  2. Confirm a gate block routes failures back to the research block

Runsight integrations and API

Runsight documentation — screenshot of the coding AI tool, Oct 2, 2026
Runsight documentation (continued). runsight.ai, captured Oct 2, 2026

Native integrations and export formats

Export: YAML

API access and separate billing

Runsight has no public API in its official documentation. CLI-based; HTTP, Python and file executors for custom tools are documented. API documentation ↗

Privacy, data use and commercial rights

Summarized from the vendor's current policies; this is not legal advice and a vendor's badge is not an independent audit.

How your inputs and outputs are handled

Self-hosted; the homepage says it runs on your machine with your keys and nothing leaves your network unless you want it to.

Best Runsight alternatives for different needs

"Better suited" is based on documented features and pricing, not on a head-to-head test.

n8n: better suited to

Zapier: better suited to

Make: better suited to

Alternatives
AlternativeWhen to choose itPricingLimitation
n8n——
Zapier——
Make——

Explore Runsight further

Best Runsight alternativesTools for the same job, by need and starting price.Best AI coding toolsOur picks in this category, compared by price and free tiers.

Tools and guides near Runsight

Runsight user reviews

No user reviews yet.

Used Runsight? Be the first to share your experience.

Write a review

Reviews are checked by a moderator before they appear. Your email is never published. Plain text only.

Your rating (required)

Show what you made or what went wrong. Images are published only if the review is approved; location data is removed.

    Runsight FAQ

    How do I start Runsight?

    Run uvx runsight; the homepage says no signup or cloud account is needed and it opens on localhost:8000. source (opens in a new tab)

    Can Runsight cap spending on an agent run?

    Yes. A budget block can set max_cost and max_tokens, and execution is killed before overspend. source (opens in a new tab)

    Which block types does Runsight support?

    Five: linear, gate, code, loop and workflow. source (opens in a new tab)

    Does Runsight send my data to a cloud?

    The homepage says it is self-hosted with your own keys and nothing leaves your network unless you choose; no privacy policy was available. source (opens in a new tab)

    Sources and review methodology

    Official sources and verification dates

    Not verified: Pricing; Privacy policy; Hosted offering; License name.

    How we documented and maintain this page

    Every fact about Runsight above comes from the 2 official pages listed here, last compared on Sep 29, 2026; we have not run Runsight ourselves, so it has no editorial score. How we check and re-check pages.

    Article updated Oct 10, 2026; prices checked Sep 29, 2026. Outbound links to vendors are plain links; we do not earn commissions. How we review · Report an error