PortCon: The AI Software Factory Summit

Unblocked vs Port

Unblocked vs Port Context Lake: how code research compares to live service state, ownership and governed actions for AI agents.

Michael Douglas
Michael Douglas
November 9, 2026
Michael Douglas
Michael Douglas&
November 16, 2026
Michael Douglas
Michael Douglas&&
November 16, 2026
Unblocked vs Port

Unblocked researches your engineering history. It reads your code, pull requests, Slack threads, tickets, and docs, and hands a coding agent a cited answer to why the code looks the way it does. Port Context Lake holds your engineering operation: what's running, who owns it, what it's worth to the business, and what an agent may change, plus the actions to change it under approval. If your agents write code that doesn't fit your codebase, Unblocked goes straight at that. If they need to take work from ticket to production, they also need what Unblocked doesn't model: live state, declared ownership, and governed actions.

This piece is written for engineers and engineering leaders working through these questions:

  • What does Unblocked do? 
  • What is Unblocked? 
  • What are the different approaches to setting up agents across the SDLC?
  • How does Port’s Context Lake compare to Unblocked? 
  • How to choose between the two? 

What does Unblocked do?

Unblocked gives coding agents and engineers the reasoning behind your code. It indexes code, pull requests, chat, tickets, and docs, works out where they contradict each other, and returns one cited answer through an MCP server, Slack, the web, a CLI, or an API. On top of that engine, it sells AI code review, a CI failure agent, and a hosted coding agent.

The problem it targets is specific. A coding agent can read your repository, but it can't see that a Slack thread three weeks ago reversed what the Confluence page still says, or that the retry behavior was settled in a PR comment rather than a doc. Unblocked's own docs describe the result: agents write code that won't merge because they don't know why the codebase looks the way it does.

Underneath is what Unblocked calls a context engine. At ingestion, it links artifacts across systems, learns that "phoenix" in Slack and "checkout-v2" in Jira are the same project, and mines recurring review feedback into team conventions. At question time, it ranks and compresses what it found, weighs recency and authority when sources disagree, flags what it can't settle, and cites the PRs, threads, and docs behind every answer.

Agents reach it through one MCP server with two read-only tools, one for research and one for fetching URLs, in Claude Code, Cursor, GitHub Copilot, Codex, and Windsurf. Since August 2026, Unblocked Code also takes a task from a Slack thread or PR comment, writes and tests the change on an isolated machine, and opens a draft PR. Unblocked's own benchmark puts the payoff at 83% faster and 48% fewer tokens on a Kotlin SDK task, and the company open-sourced the simulator so you can rerun the test on your own repo.

What is Unblocked?

Unblocked is a Vancouver-based company founded in 2021 and describes itself as "the context layer for agentic software development." It started as a Q&A tool for asking questions about a codebase in Slack or the IDE, and through 2026 rebuilt itself around agents with an overhauled MCP server, a CLI, AI code review, and Unblocked Code. It connects to more than 30 source types, from GitHub, Slack, Jira, and Confluence to SharePoint, Datadog, and Snowflake, and engineers can start asking questions the afternoon they connect a repository.

Approaches to build your agentic SDLC stack, and where Unblocked fits

Teams wiring agents into software delivery tend to take one of two paths. One pairs a research service like Unblocked with a separate connector for every system an agent writes to. The other starts with an Agentic SDLC Platform: one system of record for services, ownership, business context, and policy, with agents' actions running through it. Unblocked fits the first path directly and plugs into the second.

  • Path 1: a research service plus connectors. This is the setup Unblocked recommends. Agents call Unblocked to understand why the code looks the way it does, then call separate MCP servers for GitHub, Jira, ArgoCD, PagerDuty, and your cloud provider to make changes.
  • Advantages: you're live in a day, agent-written code matches your team's conventions sooner, and the research step costs fewer tokens than letting an agent crawl five systems on its own.
  • Trade-offs: every write goes through a different connector with its own credentials and permission model, the same security problem Unblocked's July post warns about for reads. Nothing in the path checks whether a service is tier-1, mid-incident, or under a deploy freeze before the agent acts, and because ownership is inferred from activity, an incident-routing agent has nothing authoritative to route on.
  • Path 2: an Agentic SDLC Platform underneath. Start with one model of what you run: services, deployments, environments, incidents and teams, plus properties like tier, SLA and compliance scope, reconciled continuously from GitHub, ArgoCD, PagerDuty, Datadog, Kubernetes and your cloud accounts. Agents read from that model and make changes through actions defined on it. A research service plugs in on top for the questions the model can't answer.
  • Advantages: agents make every change through the same set of actions, each with an approval step and permissions, and every run lands in one audit log. Ownership and tier are declared once, so every automation reads the same answer, and you own the schema as your org changes.

‍ade-offs: you model your domain before you see value, which takes weeks rather than a day. The model also doesn't explain why the code is the way it is, so the PR debates and Slack decisions still need a research service, or they stay in people's heads.

Where Unblocked fits: it's the research half of Path 1, and a strong research option to plug into Path 2. On either path, by its own account, it doesn't hold what's running or execute changes to it.

Port Context Lake vs Unblocked

Unblocked models engineering as a body of knowledge to research: everything your team has written, ingested, and reasoned over to explain why. Port Context Lake models engineering as a system to operate. 

Every row in the comparison below follows from that one difference.

Take an agent changing the retry logic in the payment service.  It needs both answers, and they come from different places: one from what people wrote, the other from the current state of your systems.

What Unblocked returns What Port returns
Exponential backoff rejected in PR #2317, review thread linked Tier-1 service, owned by the Payments team
Slack thread where the team settled on fixed retries, linked Open P1 incident in PagerDuty
Team convention: retry changes run the payments integration tests Production-readiness scorecard failing on SLO coverage
Cited sources for each line Deploy requires approval from a Payments lead

The difference shows up first in what each product treats as truth. When sources conflict, Unblocked grounds answers in the main branch, which is the right call for a question about code. Operations is a different question. Merging code doesn’t deploy it, and deploying it doesn’t make it healthy. The main branch can’t answer either. Port reconciles a declared model against your live systems, so "who owns this" and "is this tier-1" are facts you set rather than conclusions the system reaches.

It shows up in ownership. Unblocked pitches its expertise topology as knowing who actually knows the payment service, as opposed to who the org chart says owns it. That's the right answer to "who should I ask?" and the wrong one for "who gets paged?", because on-call is a decision your org makes, not a pattern in commits.

And it shows up in what agents can do. Unblocked's MCP server is read-only, and its own guidance sends writes to other connectors. Port's MCP server exposes the same self-service actions your developers use, each with its own approval step and permissions.

Capability Port Context Lake Unblocked
Operational Actions
Access from coding agents ✅ One MCP server with catalog data and self-service actions as tools ✅ One MCP server into Claude Code, Cursor, Copilot, Codex, Windsurf and JetBrains
Agent actions on the estate ✅ Self-service actions: deploy, roll back, provision environments, restart workloads ❌ Two read-only MCP tools; its own guidance sends writes to per-tool connectors
Approval gates at invocation ✅ Manual approval steps and RBAC on each action ◐ Relies on your code host's branch protection and reviewers; optional auto-approval for low-risk PRs
Guardrails tied to service state ✅ Block an action on scorecard level, SLA or any property ◐ Risk policies in .unblocked/risk-policies.yaml score PRs by path and criteria, with no link to service tier or production state
Automations triggered by change ✅ Workflows fire on entity events: ticket created, scorecard dropped, incident opened ◐ Triggered by a Slack mention, a PR comment or a CI failure
Live Operational State
Deployment and environment state ✅ ArgoCD, GitHub Actions and GitLab CI tied to service and environment ◐ CI runs read for failure analysis; no record of what's deployed where
Incidents and on-call ✅ PagerDuty, Opsgenie and Datadog alerts linked to the owning service ◐ Datadog and Sentry incidents, logs and errors queried on demand; no PagerDuty, Opsgenie or on-call schedule
Infrastructure and cloud ✅ AWS, GCP, Azure, Kubernetes and Terraform ◐ Google Cloud MCP (Compute, Logging, Monitoring) on demand; no AWS, Kubernetes or Terraform integration listed
Source of truth when sources conflict ✅ Declared model, reconciled against live systems ◐ The main branch, per its pricing FAQ
Freshness ✅ Continuous reconciliation from each integration ◐ PR and issue comments re-indexed in seconds; Q&A code index rebuilt every 8 hours on paid plans
Business Context
Criticality and tier ✅ A property on every service, read by agents, scorecards and actions ❌ Only if someone wrote it in a doc
SLAs and SLOs ✅ Modeled per service and linked to incidents ❌ Not modeled
Compliance scope ✅ A property that can gate actions and scorecards (for example PCI or SOC 2 scope) ❌ Not modeled
Cost ✅ Cloud cost attributed to services through cloud integrations ❌ Not modeled
Product and domain mapping ✅ Services related to products and domains through blueprint relations ◐ PostHog analytics and Zendesk tickets readable as documents, not linked to services
Ownership and Governance
Ownership ✅ Declared team-to-service relations ◐ "Expertise topology" inferred from PR reviews, Slack answers and Jira activity, weighted by recency
Permission enforcement ✅ RBAC on entities and actions ◐ Data Shield off by default for every source except Slack; Datadog, Sentry, Snowflake, Stack Overflow and websites not covered
Standards enforcement ✅ Scorecards evaluate every service against rules you define ◐ Conventions mined from review feedback, enforced on pull requests rather than running services
Audit trail ✅ Action log with actor, inputs and outcome ✅ Audit logs, plus source citations on every answer
Ontology and Schema
Define your own entity types ✅ Blueprints for any entity, property and relation ❌ Internal knowledge graph of artifacts (PRs, tickets, threads, code); not customer-defined
Custom and proprietary sources ✅ Custom integrations use the same API as native ones ◐ Custom MCP (read-only tools) and API document collections, capped at 25 collections
Portability ✅ Terraform and Pulumi definitions, full API export ❌ Graph is internal, with no export
Code Understanding (where Unblocked leads)
PR discussion, chat and docs ❌ Structured entities only; discussion text and docs not indexed ✅ 30+ source types indexed, with links back to each original
Contradiction resolution ❌ Not attempted ✅ Weighs code behavior, recency and authority; flags what it can't settle
Ranking and compression ❌ Entity data returned as stored ✅ Server-side ranking and compression, with an effort setting
Team conventions ❌ Not modeled ✅ Mined from recurring review feedback, scoped to repos and teams
Memory across sessions ❌ Agent sessions start without memory ✅ Shared memories checked against sources before they're saved
AI code review ❌ Not offered ✅ Inline review, risk labels and merge-gating status checks
Hosted coding agent ◐ Routes tickets to Cursor, Claude Code or Copilot with service context and an approval step ✅ Unblocked Code: Slack or PR comment in, verified draft PR out

Bottom line: Unblocked helps agents understand your code, and Port gives them the live state, ownership, and governed actions they need to operate your systems.

How to choose, and where to start

If your agents mostly write code and the problem is that it doesn't fit your codebase, start with Unblocked. If your agents need to take work from ticket to production, route incidents or change infrastructure, they need an Agentic SDLC Platform underneath, because something has to hold what's running and govern what agents may change. Unblocked earns its place on the first job, because the context it pulls from years of PR reviews and Slack decisions is something a model can't reconstruct on its own.

On the second job, if we take Port’s customer dLocal as an example, they show what having live state, ownership, and guardrails underneath looks like. dLocal runs an internal coding agent, dCoder, that takes 45% of its engineering tickets from start to finish, from reading the ticket to opening a reviewed pull request. Teams using it perform more than 20% better, which dLocal equates to a 20% larger engineering team with no added headcount. dLocal didn't start with the agent: it put Port first as the source of truth for its services, infrastructure, repositories, teams, and policies, then built dCoder on top, so the agent runs the routine work inside the context and guardrails Port holds. As CTO Alberto Almeida put it, "My entry point for all of that is Port."

FAQ

What does Unblocked do?

Unblocked indexes your code, pull requests, chat, tickets and docs, reconciles where they disagree, and gives coding agents and engineers a cited answer to why your code looks the way it does. It also sells AI code review, a CI failure agent and Unblocked Code, a hosted coding agent that opens draft pull requests.

What is an Agentic SDLC Platform?

An Agentic SDLC Platform gives AI agents what they need to work across the software lifecycle: context on what's being built and what's running, ownership and policy covering who's accountable and what's permitted, and governed actions to make changes, with humans approving at the points that matter.

Is Unblocked an Agentic SDLC Platform?

No. Unblocked is a research service for engineering knowledge, which it describes as the research layer agents call to understand why something was built the way it was. It doesn't model live operational state, business context or declared ownership, and its agent-facing tools are read-only.

What are alternatives to Unblocked?

For AI code review, Unblocked compares itself with Greptile, and CodeRabbit covers similar ground. For company-wide knowledge search, Glean is the usual comparison. For operational context and governed agent actions across your engineering stack, Port is an Agentic SDLC Platform.

Unblocked vs Port Context Lake: which is right for me?

Choose Unblocked if your main problem is agent-written code that doesn't fit your codebase. Choose Port if your agents need to deploy, roll back, provision environments or route incidents, and need the live service state, declared ownership and business context to do it safely across GitHub, ArgoCD, PagerDuty, Datadog, Kubernetes and your cloud. Many teams will want both.

When should I choose Unblocked?

Choose it if your agents mostly write code, rework and review cycles are the cost you're tracking, your team's decisions live in PR threads and Slack, and you want something useful on day one without modeling your domain.

Can I use Unblocked and Port together?

Yes. Unblocked's Custom MCP integration connects any HTTPS MCP server as a read-only source, so Port's MCP server can feed it live service, ownership and incident data. Your coding agents get the reasoning behind the code from Unblocked and make changes through Port's actions.

Tags:
{{download_survey}}

Download the RFP guide

Thanks for requesting the RFP Ebook
You can download it here
Thank you!You’ll be notified when the guide hits!
{{download_report}}

Download the report

Thanks for submitting.
Check your inbox for the report download.
Thank you!You’ll be notified when the guide hits!
{{survey-buttons}}

Get your survey template today

By clicking this button, you agree to our Terms of Use and Privacy Policy
{{stay_tuned}}

Stay tuned for our upcoming tutorial

With a step-by-step guide walking you through how to implement and scale Anthropic’s playbook in Port’s free tier. Register to be notified once the guide is published:

By clicking this button, you agree to our Terms of Use and Privacy Policy
Thank you!You’ll be notified when the guide hits!
{{survey}}

Download your survey template today

By clicking this button, you agree to our Terms of Use and Privacy Policy
{{roadmap}}

Free Roadmap planner for Platform Engineering teams

  • Set Clear Goals for Your Portal

  • Define Features and Milestones

  • Stay Aligned and Keep Moving Forward

{{rfp}}

Free RFP template for Internal Developer Portal

Creating an RFP for an internal developer portal doesn’t have to be complex. Our template gives you a streamlined path to start strong and ensure you’re covering all the key details.

{{ai_jq}}

Leverage AI to generate optimized JQ commands

test them in real-time, and refine your approach instantly. This powerful tool lets you experiment, troubleshoot, and fine-tune your queries—taking your development workflow to the next level.

{{cta_1}}

Check out Port's pre-populated demo and see what it's all about.

Check live demo

No email required

{{cta_webinar_aug_18}}

LIVE WEBINAR, Aug 18, 2026:

Context-aware Vibe Coding for Platform Engineering

{{cta_webinar_oct_22}}

Thursday, October 22 12:00pm EDT⋅6:00pm CET

To learn more about the new capability and to see a live demo, join our upcoming community session with GitHub

{{cta_explore_port}}

Move fast while staying in control

Build governed agentic workflows on one central platform.

{{public_demo}}

See it in action:

Watch this video on generating Terraform with Port, or explore our public demo.

{{cta_survey}}

Check out the 2025 State of Internal Developer Portals report

See the full report

No email required

{{cta_2}}

Minimize engineering chaos. Port serves as one central platform for all your needs.

Explore Port
{{cta_3}}

Act on every part of your SDLC in Port.

Schedule a demo
{{cta_4}}

Your team needs the right info at the right time. With Port's software catalog, they'll have it.

{{cta_5}}

Learn more about Port's agentic engineering platform

Read the launch blog

Let’s start
{{cta_6}}

Contact sales for a technical walkthrough of Port

Let’s start
{{cta_7}}

Every team is different. Port lets you design a developer experience that truly fits your org.

{{cta_8}}

As your org grows, so does complexity. Port scales your catalog, orchestration, and workflows seamlessly.

{{cta_n8n}}

Port × n8n Boost AI Workflows with Context, Guardrails, and Control

{{port_builders_session}}

Port Builders Session: A Single, Governed Interface for All MCP Servers

{{cta-demo}}
{{read_case}}
{{n8n-template-gallery}}

n8n + Port templates you can use today

walkthrough of ready-to-use workflows you can clone

Template gallery
{{from_manual_to_autonomous_engineering}}

From manual to autonomous engineering

One platform to build, govern, and operate the Agentic SDLC.

Explore Port
{{port_is_open_for_you_to_try_it}}

Port is open for you to try it

build your first agentic workflow today

Sign up
{{reading-box-backstage-vs-port}}
{{cta-backstage-docs-button}}

Starting with Port is simple, fast, and free.