TL;DR: We've spent years solving the context problem for engineers. Now we're solving it for agents.
Four years ago, we set out to solve a problem that frustrates every developer I know: finding the right information to get work done is a chore. A team’s knowledge is spread across Jira tickets, Slack threads, Confluence docs, the codebase, and the heads of experienced colleagues. Getting that information is tedious, and reconciling it adds hours to every project.
To address this, we built product experiences that tens of thousands of developers use every day: a web dashboard for asking questions, deep integrations with Slack, and a code review tool that weighs in like the best engineer on the team.
Underneath all of this is an engineering-focused context engine, a system that gathers, organizes, and retrieves the information needed to answer a question, then produces a grounded response.
Over the past few months, usage has (materially) shifted.
More of the questions are coming from agents, and they need that same tribal knowledge to accomplish their tasks effectively.
Teams are using coding agents to handle planning, implementation, code review, and incident investigation. They’re also building workflows to enrich support tickets, validate bugs, and improve CI pipelines.
The problem is the same: context.
Agents are still searching across fragmented systems and settling for answers that look plausible.
This wasn’t hard to see coming.
We launched a beta of Unblocked MCP months ago and adoption was immediate.
Today, we’re taking it out of beta and into general availability, alongside a set of improvements and new tools driven directly by customer feedback.
Unblocked across your agent ecosystem#
No matter how you're using agents, and whether you choose Cursor, Claude, Codex, or Copilot, we've made sure Unblocked will fit into your workflow.
We’re releasing the fully overhauled Unblocked MCP server. Over the last several months, our customers have pressure tested the MCP in a variety of use cases, through intentional prompting and custom skills. Their feedback led to today’s release, which dramatically improves performance and answer quality and optimizes for how often the tool gets called - making sure we show up with the right answers at the right time.
We’re also releasing a CLI for deterministic, scriptable access to Unblocked. It’s designed for CI, scripts, and background jobs, and can be used directly by agents when you want explicit, reliable control over how context is pulled in.
Prebuilt skills for common agent tasks#
To further support agentic workflows, we’re releasing a cookbook of prebuilt skills using the CLI and MCP that you can fork and customize. Each one helps your agent gather the context it needs to complete real tasks like planning, code review, ticket enrichment, and bug validation. Every skill also includes a headless version for running in CI, scripts, or autonomous agents.
From prompt to PR, fully autonomous#
If you’re building or running autonomous agents, we’ve put together a tool that shows how Unblocked brings institutional memory into Claude-managed agents. It clones a repo, gathers context through Unblocked, implements changes, and opens a PR, grounded in your organization’s code, conversations, docs and issues.
The difference it makes#
Context means something different to everyone you ask. But our customers are clear on the benefits they get from using Unblocked to bring it into their workflows. Users have consistently reported significant time savings and quality of life improvements. When they start using Unblocked with agents, those numbers multiply quickly. It's not just time savings; it's less baby sitting with fewer wasted tokens and rework loops.
Here’s what customers have to say:
"The agent asks Unblocked for everything — the ticket, the Slack context, what's been done in related repos — and then it starts implementing. It's especially powerful for cross-repository work where you'd otherwise have to do all that archaeology yourself." — Arthur Rodolfo at Clio
"My setup tells the agent: before you implement anything, go check Unblocked. It has everything — our repos, Notion, Slack, coding standards — and it surfaces things I wouldn't have thought to look for." — Justin McCraw at The Information
"You cannot make coding agents work without domain and functional context. We connected Unblocked to our repos, Atlassian, internal docs, product documentation, and Slack history. When an agent asks a question, it gets the full picture — not just the code analysis, but also why decisions were made and what the constraints are. Other tools like Copilot know only the code. That's limited value." — Raphael Bres, CTO, Tradeshift
"LLMs are only as good as their context, and code alone gives them a narrow view — the codebase is always behind where the team wants to go. When I plugged Unblocked into our context-gathering and PR review steps, it brought in the Slack conversations where real architectural decisions get made. We went from three rounds of PR review to one before the code was production-ready." — Pablo Vallejo, Engineering Manager, Clio
"During a security incident I asked Unblocked to find every repository we had that could be vulnerable and report back. It came back with repositories I wasn't fully aware of. That's the pattern — I know what I'm looking for, but Unblocked finds what I don't know about." — Youssef Eladawy at HeyJobs
"My agents know to use Unblocked first because it's faster than having them figure it out from the code alone. Individual MCPs are great when you already know what you're looking for. Unblocked is what you use when you need the full picture." — Zachary Goldberg, Engineering Manager, Lilt
"Unblocked is the first MCP queried for everything we look up. It's not just checking the code — the code could be wrong. It pulls the Confluence docs, the feature planning documents, the Slack conversations. For an engineering manager trying to understand a platform fast, that's the difference."— Sam Younger, Engineering Manager, UserTesting
"I don't even think about Unblocked doing the heavy lifting. I just get all the relevant information. That's it."— Andrei Antanovich, Engineer, Waste Logics
In our own codebase, we ran an experiment giving the same model the same task: introduce “adaptive thinking mode” into a Kotlin SDK monorepo while maintaining backwards compatibility across three modules.
- Without Unblocked context, it took two and a half hours, 20.9 million tokens, and four rounds of human correction. And it still resulted in twelve compilation failures.
- With Unblocked built into the workflow, it took 25 minutes, less than half the tokens, required zero human corrections, and resulted in zero failures.
We built a harness so you can run the same experiment against your own codebase — measuring time, token usage, output quality, and failure rate, with Anthropic or OpenAI models. If you're still not convinced, try it and see.
Get started#
If you're already using Unblocked, getting started with agents is straightforward. The docs cover setup for the MCP, CLI, and Claude Desktop connector. You can grab a recipe from the Cookbook to fork and adapt to your setup.
If you're new to Unblocked, we'd love to show you how we can save your team time and vastly improve agent output.
Book a demo and we'll walk you through it.
