← Back to Carapace

Carapace

The control layer that governs how agents execute tasks, access resources, and coordinate work.

What is Carapace?

Carapace is a work management dashboard for teams that include AI agents. It shows you what work is happening, who is doing it, and why it was started — all stored as linked data on a W3C Solid Pod that you control.

Unlike Jira or Linear, Carapace is designed for a world where autonomous agents create tasks, update statuses, and coordinate work alongside humans. The knowledge base is a shared graph that both humans and agents can read and write.

Key principle: Your data lives in your Solid Pod, not on our servers. Carapace is a viewer and editor — it connects to your Pod, loads the data into your browser, and never sends it anywhere else.

Getting Started

1. Create an account and Pod on crawlout.io

Carapace stores your work data on a Solid Pod. This alpha version uses crawlout.io (operated by Interition). You need three things: an account, a Pod, and a WebID. Here's how:

  1. Go to crawlout.io account registration
  2. Register with an email and password — this creates your account
  3. After registering, you'll see your account page. Click Create pod — give it a short name (e.g. alice). This creates both your Pod (at https://crawlout.io/alice/) and your WebID (at https://crawlout.io/alice/profile/card#me) automatically
Why three steps? Community Solid Server separates accounts, Pods, and WebIDs. This is confusing but intentional — one account can have multiple Pods and WebIDs. For most users, you'll create one of each and be done.

2. Login

Click Login to crawlout.io in the top right. You'll be redirected to the crawlout.io login page. Enter the email and password you registered with. After authenticating, authorise Carapace when prompted, and you'll be redirected back.

Your session is saved in your browser — you won't need to login again until your session expires or you click Logout.

3. Set up your Knowledge Base

After logging in, if you don't have a knowledge base yet, click Setup KB in the top right. The setup page will:

Setup takes a few seconds. Once complete, you'll be redirected to the dashboard with your data loaded.

Alternatively, if you prefer to set up the KB manually or via an agent, the SKILL.md includes command-line scripts for creating the structure.

4. Enter your Knowledge Base URL

If you already have a knowledge base, enter its URL in the Knowledge Base field — for example https://crawlout.io/team/work/ — and click Load. The URL is remembered for next time.

Reading the Dashboard

Stats bar

The top bar shows counts at a glance: P1 items that need attention, work in progress, open items waiting to start, parked items, and completed work.

Filters

Click a filter to show only work items matching that status. P1 Only shows all statuses but only critical priority items.

Triggers

Triggers answer "why does this work exist?" Each trigger is an event — a strategic decision, a competitor move, an industry announcement, or an internal discovery. Work items link back to triggers so you can always trace from a task to the reason it was created.

Epics and Tasks

Epics are strategic themes. Tasks are grouped under epics. Each task shows:

Tasks without an epic appear in an Unlinked section — this is a signal that work is happening without a strategic reason. You can link them to an epic from the dashboard.

Agent Workload

A bar chart showing how many tasks each agent (or human) owns. Useful for spotting overloaded agents or unbalanced workloads.

SPARQL Query

For advanced users. The query panel lets you run SPARQL queries directly against the knowledge base loaded in your browser. Preset queries cover common questions like "what's active?", "what has no trigger?", and "show dependencies". You can write your own queries using the ontology.

Creating Work Items

Add Trigger

Click + Add Trigger in the Triggers section. Choose a type (Strategic Objective, Competitor Action, Industry Event, Internal Discovery, or Customer Signal), give it a name and date, and optionally add a description and evidence URL. The trigger is saved directly to your Pod.

Add Epic

Click + Add Epic to create a strategic theme. Link it to a trigger to maintain traceability.

Add Task

Click + Task on any epic to create a task within that epic. Tasks get an auto-generated ID and can be assigned to any agent in your team.

Link to Epic

Unlinked tasks have a Link to Epic button. Click it to associate the task with a strategic theme.

Connecting Your Agents

This is the key step. Carapace is a human dashboard. To make it useful, your AI agents need to read and write the same knowledge base. Give them the SKILL.md file.

SKILL.md is the agent skill prompt. It teaches any AI agent — Claude Code, OpenClaw, or similar — how to:

When an agent creates or updates a work item via the Solid Protocol, it appears in Carapace on the next page load. When you create a work item in Carapace, agents can read it immediately. The Pod is the shared source of truth.

How agents get access

Each agent needs a WebID on your Solid server and credentials to authenticate. The Pod owner grants read/write access to the work graph containers using Solid access control (WAC). See SKILL.md for the authentication details.

Technical Details

Carapace is a static web application with no backend. It uses:

The full ontology is in vocab.ttl.