Harness
It coordinates models, tools and company context, with each agent’s instructions and tools kept in your repository. Self-host it or run it on ours.
const agent = await adapt.agents.create({
instructions: './agents/triage.md',
tools: ['github', 'linear', 'sentry'],
})
await agent.run('Triage the new error reports')
Read the 3 new Sentry issues
Find the change behind each
File a summary in LinearUse cases
Version instructions, tools and configuration in your repository, and review changes like any other code.
Give the agent integrations, APIs and command-line tools with the access each task needs.
Use your own models, sandboxes and storage or Adapt’s, without rewriting the agent.
How it works
Give the agent a bounded job first, and add more responsibility once your team has reviewed its results.
Connect issue tracking, error reports and the relevant repository with the permissions the task needs.
Ask it to gather the error context, find related changes and draft a triage summary for an engineer.
Let engineers request the same investigation from Slack or the web, and keep useful findings for the next report.
import { Adapt } from '@adapt/sdk'
const agent = await new Adapt().agents.create({
instructions: './agents/triage.md',
tools: ['github', 'linear', 'sentry'],
})
await agent.run('Triage the new error reports')Pricing
The harness is open source. Run it on your own infrastructure at no cost, or pair it with Adapt’s primitives.
Works with
The company agent combines all four with your tools, shared knowledge and the places your team already works.