Built-in embeddingsConnect to AI

Where SEOs and agents
work together

for MacDownload for Mac for WindowsDownload for Windows

A workspace for modern SEOs. Silkra is a local spider with a built-in embedding model. Map sites by meaning, chat with your crawl, act on the data, and connect it to your AI tools.

How it works

Watch demo

Collect evidence
Scrape websites locally.
Map by meaning
Group pages by topic, not URL.
Put agents to work
Analyze, create, and monitor.

Ask your crawl like a teammate.

Drag any page into the chat and ask what you'd ask a colleague. Answers come back grounded in the embedded chunks, with the rows behind them still on screen.

Explore the workspace

Can you help analyze this page?

Reviewed link architecture

Reviewed page details

Found related pages

Reviewed link architecture

This page returns 200 and is indexable. It sits in the Billing documentation group alongside five other pages.

It overlaps closely with /docs/billing/overages — both explain proration, so they read as near-duplicates.
Nothing under /pricing links to it, which is where readers comparing plans tend to start.

Ask about your data...

/docs/billing/plan-limits

LatestPlan limits brief

Search by meaning, or type / to filter…K

Address	Status	Indexability	Semantic Group	Crawled
example.com/account/billing	200	Noindex	Account settings	2h ago
example.com/account/team	200	Noindex	Account settings	2h ago
example.com/account/usage	200	Noindex	Account settings	2h ago
example.com/docs/billing/invoices	200	Indexable	Billing documentation	2h ago
example.com/docs/billing/overages	200	Indexable	Billing documentation	2h ago
example.com/docs/billing/plan-limits	200	Indexable	Billing documentation	2h ago
example.com/docs/billing/proration	200	Indexable	Billing documentation	3h ago
example.com/docs/billing/refunds	200	Indexable	Billing documentation	3h ago
example.com/docs/billing/seats	404	Error	Billing documentation	3h ago
example.com/blog/annual-vs-monthly	200	Indexable	Editorial resources	3h ago
example.com/blog/choosing-a-plan	200	Indexable	Editorial resources	3h ago
example.com/blog/usage-based-pricing	200	Indexable	Editorial resources	3h ago
example.com/compare/plans	301	Redirect	Pricing and plans	3h ago
example.com/pricing	200	Indexable	Pricing and plans	3h ago
example.com/pricing/enterprise	200	Indexable	Pricing and plans	4h ago
example.com/pricing/nonprofit	200	Indexable	Pricing and plans	4h ago
example.com/faq/accounts	200	Indexable	Support answers	4h ago
example.com/faq/billing	200	Indexable	Support answers	4h ago
example.com/support/contact	200	Indexable	Support answers	4h ago

Latest

Search by meaning, or type / to filter…K

Issues7Info5

Broken Page (4xx Error)14Canonical Points Elsewhere9Content Over 2 Years Old58Thin Content62AI Crawlers Blocked41Images Missing Alt Text33Exact Duplicate Content21

10 groups • 228 pagesUngrouped9GroupsLinks

Billing documentation54

Pricing and plans40

Account settings33

Editorial resources26

Integration guides17

Support answers21

Onboarding basics10

Release notes13

Compliance and security8

Isolated page

No other page in Integration guides links to it, so readers and crawlers only reach it from outside the topic.

See the whole site by topic.

Every page is a tile, grouped by what it actually says. Cluster sizes, thin coverage, and the corners nothing links to are all visible before you filter anything.

See how the map works

The audit writes itself. You edit it.

Ask for the write-up and it drafts beside the chat, citing the pages it came from. Keep talking to reshape it — add a section, cut a finding, tighten the recommendation.

See how the agent works

Can you generate a quick audit?

Created document · Semantic audit

Semantic audit

Document

DownloadExport to Google Drive

It's open beside this. I grouped the 237 pages into the ten topics the clustering found, then went through each one for depth, overlap, and whether anything links to it from inside its own topic. Billing documentation and Editorial resources are where the findings concentrate — the first for reachability, the second for thin pages.

Want me to take it further?

List the nineteen isolated pages with the topic each one belongs to
Add the full similarity pairs above 0.9, not just the clearest three
Cut it down to a one-page version for a weekly update

Ask about your data...

TableSemantic audit

Semantic audit — example.com

237 pages were crawled and grouped into ten topics by what they actually say, rather than by where they sit in the URL structure. This write-up covers how coverage is distributed across those topics, which pages say the same thing as each other, and which corners of the site nothing links to from inside their own topic.

Summary

Billing documentation is the largest topic at 54 pages, and also holds the most pages nothing links to internally.
Editorial resources has the highest share of thin pages: 11 of 26 fall under the word-count threshold.
Nineteen pages across the site have no inbound link from another page in the same topic.

Coverage by topic

The five largest topics account for 174 of the 237 pages. Thin and isolated counts are shown alongside so depth and reachability can be read together.

Topic	Pages	Thin	Isolated
Billing documentation	54	6	7
Pricing and plans	40	3	2
Account settings	33	4	3
Editorial resources	26	11	5
Support answers	21	2	2

Overlapping pages

Seven pairs sit above 0.9 cosine similarity, which usually means two pages were written to answer the same question at different times. The clearest cases:

/docs/billing/proration and /docs/billing/overages both explain proration end to end, with different examples.
Three plan-comparison posts under /blog cover the same ground as /pricing and rank against it.
/faq/billing restates six answers that already exist as their own pages under /docs/billing.

Internal linking

Nineteen pages have no inbound link from inside their own topic. Most sit under Billing documentation, which readers arrive at from /pricing — so the pages exist, but the path to them runs through search rather than through the site.

Export to Google Drive

Change log

# How proration works

When you switch plans mid-cycle, we adjust your bill.

When you change plans mid-cycle, we charge or credit the difference for the days remaining in the billing period.

The adjustment appears on your [next invoice](/docs/billing/invoices) as a separate line item.

## Changing plans

## Upgrades

Upgrades start on your next renewal date.

Upgrading takes effect immediately.

Prorated amounts are calculated daily.

We bill the difference at the end of the cycle.

We bill the prorated difference between the two plans for the rest of the cycle.

## Cancellations

## Downgrades

Contact support to downgrade a plan.

Downgrades apply at the end of the current cycle, and unused credit rolls forward to the next invoice.

Downgrades below the free tier close the account.

Downgrading to the free tier keeps the workspace open.

## Refunds

## Credits

Credits apply to the next invoice automatically.

Credits never expire while the plan is active.

Page inspector

How proration works

When you change plans mid-cycle, we charge or credit the difference for the days remaining in the billing period. The adjustment appears on your next invoicehttps://example.com/docs/billing/invoicesClosely related · 0.84 cosine to this page as a separate line item.

Upgrades

Upgrading takes effect immediately. We bill the prorated difference between the two plans for the rest of the cycle.

Downgrades

Downgrades apply at the end of the current cycle, and unused credit rolls forward to the next invoice.

Every page rendered, and tracked.

Each crawl keeps the text a page actually rendered, then compares it to the last one line by line. Read the page as it stands, or see exactly what changed since you looked.

See where the evidence comes from

More ways to work the crawl.

Search the site like a model
Query real passages with cosine similarity, then walk back to the page they came from.
Explore retrieval
Hand the crawl to any agent
MCP syncs the same workspace to Claude, Cursor, and ChatGPT without another export.
See agents and MCP
Used Silkra
List Silkra Sites
Get Site Synopsis
Get Issue Detail
Billing documentation is the largest group. Three pages explain proration, and none of them are linked from /pricing — so the path to those docs runs through search rather than the site.
Compare workspaces with an agent
Attach another scrape and let the agent compare you to a competitor.
Compare scrapes
example.org
Used Silkra
Aggregated compare workspace
Queried compare workspace
example.org has a live /docs/billing/seats page. example.com still 404s there, and /pricing does not link into the billing docs on either site.

The checks that run themselves.

Find pages leaking authority
Noindexed URLs collecting internal links, orphans, and hubs that pass equity nowhere.
Split pages the way models do
Every page is chunked and embedded, so you read the passage an assistant would pull.
Extract the signals on a page
Tables, code blocks, quotes, and statistics come out of the render as filterable fields.
Read the link graph against meaning
Compare what links to what with how closely the two pages actually relate.
Catch thin and redundant pages
Low-substance URLs and two pages competing for one intent, flagged by similarity.
Clear the foundations
Status codes, canonicals, indexability, and metadata hygiene in a single pass.
Prioritize by page value
Sort findings by the value of the pages they touch instead of by raw issue counts.
Group work by template and topic
Fix the template once rather than walking nine hundred URLs one row at a time.
See what changed since last crawl
Re-scrape and every page keeps its first-seen and content-changed history.

The audit, minus the busywork.

Get started

without silkra

with silkra

Pivoting a 50,000-row exportAsk the crawl a question
A notebook to get cosine similarityClusters the moment the crawl ends
Topics guessed from URL foldersTopics read from the actual words
Snippets pasted into ChatGPTThe whole workspace over MCP
Audit decks rebuilt by handDrafts from evidence already attached

Give agents SEO evidence, not slop

Crawl, chunk, and embed any site on your machine, then hand your agents the passages and page evidence behind every answer.

for Mac for Windows

free to start

Cosine	Address
0.74	/docs/billing/proration
0.71	/docs/billing/invoices
0.68	/docs/billing/overages
0.64	/pricing
0.61	/faq/billing

Where SEOs and agentswork together

How it works

Collect evidence

Map by meaning

Put agents to work

Ask your crawl like a teammate.

See the whole site by topic.

The audit writes itself. You edit it.

Every page rendered, and tracked.

More ways to work the crawl.

Search the site like a model

Hand the crawl to any agent

Compare workspaces with an agent

The checks that run themselves.

Find pages leaking authority

Split pages the way models do

Extract the signals on a page

Read the link graph against meaning

Catch thin and redundant pages

Clear the foundations

Prioritize by page value

Group work by template and topic

See what changed since last crawl

The audit, minus the busywork.

Give agents SEO evidence, not slop

Where SEOs and agents
work together