Specly — the bridge between people and the code that builds the data platform

What Specly is

A specification your platform is actually built from.

Wikis go stale, spreadsheets get forgotten, and the real definitions end up in people's heads. Specly keeps the model, the documentation and the intent in one place — versioned, searchable, and shipped with the code.

◈

Design before you build

Sketch facts and dimensions on a canvas, agree on the model with the business, and mark what is a draft versus what is approved for development.

≡

Document while it matters

Meanings, owners, data types, naming rules, sensitivity and lineage captured at design time — not reconstructed a year later.

⇄

Ship the spec with the code

Every schema lives as files in your Git repo, in sync both ways. The specification travels in the same release as the transformation that implements it.

Specly project overview dashboard showing diagram, table, data product and source system counts — The whole platform at a glance: diagrams, tables, data products, status.

Shared understanding

Everyone reads the same model — in their own language.

Business defines what a term means. Engineering sees what to build. Nobody has to translate a meeting into a table by hand.

✓ From a business matrix to a diagram — start from the conversation, end with facts, dimensions and relations everyone recognises.

✓ Draft or approved — a clear line between what is still a sketch and what is ready for development.

✓ Who to ask — every table and data product carries its owner, so a question takes one click, not three Slack threads.

✓ Consistent by construction — naming standards, data types and required documentation enforced as rules, not as review comments.

Specly diagram view showing a star-schema data model with fact and dimension tables connected by crow's-foot relations — The full model, hand-arranged and zoomed to fit — facts, dimensions and the crow's-foot relations between them.

One place, many questions

The questions your platform can finally answer.

The same specification answers very different people — because it holds the meaning, not just the DDL.

“Which changes are approved for development, and is the spec complete enough to build?”Data engineer

“Where is this column used downstream, and what breaks if I change it?”Platform engineer

“Do we agree on what active customer means before we model it?”Business analyst

“Which tables hold sensitive data, and who is it shared with?”Governance

“Which data product is 90% built already — what should we do next?”Product lead

“Table X failed overnight. Who do I notify, and how urgent is it?”Platform operations

Build with AI

Knowledge your agents can build from.

Specly is the context layer between what your organization means and what your agents write. Schemas as files, exposed over MCP — so the model in the tool and the code in the repo never drift apart.

✓ Files in your repo — every table is a schema file, synced both ways, reviewed like any other change.

✓ Work where you work — edit from your editor and your agent, and keep Specly current without leaving it.

✓ Verify what was built — compare the live platform against the design and see where they disagree.

specly/tables/dim_customer.yaml

# synced both ways with your repo
name: dim_customer
layer: gold
owner: customer-domain
status: approved_for_development
description: One row per customer, current state.
columns:
  - name: customer_key
    type: bigint
    primary_key: true
  - name: is_active
    type: boolean
    description: Purchase in the last 12 months.
    sensitivity: internal

Yours to shape

Configurable to how your team works — opinionated about the rest.

Define your own table types, custom fields, statuses, layers and naming rules. The principles that keep a platform coherent are built in; the vocabulary is yours.

Admin configuration screen for table types and custom data product metadata fields

The bridge between people and the code that builds the data platform.