The SIGNLD extension for Claude: FAQ

The SIGNLD extension for Claude puts a company's Knowledge Graph inside a Claude session so answers reference real entities, metrics and records instead of general knowledge alone. This page collects the questions teams ask most often about what the extension does, what it does not do, and how it stays read-only and traceable.

By SIGNLD Editorial · · 8 min read · Product explained
The SIGNLD extension for Claude: FAQ

Related reading in this series: Permissions and data access in the SIGNLD extension for Claude and Setting up the SIGNLD extension for Claude.

In this article

What is the SIGNLD extension for Claude?

It is an extension that gives a Claude session access to a company's SIGNLD Knowledge Graph, the private map of entities, metrics, rules and relationships SIGNLD builds from a business's connected systems. Instead of answering from general training data alone, Claude can reference that graph inside the conversation. The extension does not change what Claude is; it changes what business context is available to it during a session.

Who builds and maintains the extension?

The SIGNLD extension for Claude is built and maintained by SIGNLD using Anthropic's publicly available extension interfaces. SIGNLD is not affiliated with, sponsored by, or endorsed by Anthropic. SIGNLD owns the code, the Knowledge Graph connection, and the support relationship for the extension. Anthropic owns Claude itself and the interfaces the extension is built against.

Does Anthropic build or endorse the extension?

No. Anthropic did not build the SIGNLD extension for Claude and does not endorse, sponsor, or partner on it. SIGNLD built it using Anthropic's publicly available extension interfaces, the same kind of interface any developer can build against. Any Claude feature, limit, plan or model detail belongs to Anthropic and is documented at docs.anthropic.com, because those details change independently of SIGNLD.

Can the extension write to my source systems?

No. The extension, like every part of SIGNLD, holds read-only connections to source systems. It can retrieve and reference records through the Knowledge Graph, but it cannot create, edit or delete anything in a connected CRM, ERP, spreadsheet or other system. Nothing a user asks Claude through the extension results in a write back to a source system. Read more on the security page.

What does the extension actually add to a Claude session?

It adds access to the Knowledge Graph: named entities, defined metrics, business rules and the relationships between them, all sourced from systems the company has connected to SIGNLD. A question Claude answers with the extension active can reference a specific customer, a specific metric definition, or a specific record, rather than answering only from general reasoning. See concepts for how the graph itself is structured.

How does an answer stay traceable to a source record?

Every piece of Knowledge Graph context the extension surfaces links back to the source record it came from, whether that is a row in a spreadsheet, a deal in a CRM, or a line in a financial system. The person asking can follow that reference back to where it originated instead of taking the answer on faith. This traceability applies inside a Claude session the same way it applies in SIGNLD's own interface.

What is the Knowledge Graph the extension draws from?

It is the private structure SIGNLD builds by connecting to a company's business systems and mapping the entities, metrics, rules and relationships those systems contain. It is unique to each SIGNLD tenant and is not shared across customers. The extension does not introduce a separate graph; it gives Claude access to the same graph SIGNLD's own interface already uses. Full detail on how the graph is built is in concepts.

Do I need a SIGNLD account to use the extension?

Yes. The extension connects a Claude session to a specific SIGNLD tenant's Knowledge Graph, so a SIGNLD account with connected systems has to exist first. Without an account and at least one connected system, there is no graph for the extension to reference, and Claude behaves as it normally would without it.

What systems can the Knowledge Graph connect to?

SIGNLD connects to business systems a company already runs, including spreadsheets, CRMs, ERPs and other operational tools, through read-only connections. The extension itself does not add or remove which systems SIGNLD can connect to; it simply gives Claude access to whatever Knowledge Graph a tenant has already built from its connected systems. Adding a new source system to SIGNLD makes that context available through the extension as well.

Can I control what Claude can see through the extension?

Access follows the same permission model as the rest of SIGNLD. A user's role and permissions determine which parts of the Knowledge Graph they can query, and the extension respects those same boundaries inside a Claude session. Nobody gains broader access to source data by using Claude instead of SIGNLD's own interface; the underlying permission rules stay the same.

Where is the AI inference for SIGNLD's side of this running?

SIGNLD's own AI processing runs on a single-tenant AWS Bedrock instance dedicated to that tenant, under SOC 2 Type II controls, with tenant isolation enforced between customers, and it is never trained on a customer's data. This applies to how SIGNLD prepares and serves Knowledge Graph context to the extension. Claude's own model processing is Anthropic's infrastructure, documented separately at docs.anthropic.com.

Does the extension change Claude's pricing or model limits?

SIGNLD does not set or change Claude's pricing, usage limits or model behavior; those are Anthropic's terms and they change over time, so they are documented at docs.anthropic.com rather than restated here. The extension is a separate layer that adds Knowledge Graph context to a session; it does not alter what Claude itself charges for or how its models are governed.

What happens if the Knowledge Graph has not seen a connected system yet?

Claude answers from general reasoning alone for anything the Knowledge Graph has no structure for, the same as it would without the extension. The extension does not fabricate business context to fill a gap; it surfaces only what the graph actually contains and links to a source record. A missing connection means missing context, not a guessed answer presented as grounded.

Is the extension a replacement for SIGNLD's own interface?

No. It is an additional place to reach the same Knowledge Graph, useful when someone is already working inside a Claude session and wants business context without switching tools. SIGNLD's own interface still produces the full Decision Brief format, with a finding, evidence, a confidence score and a recommended action, and remains the primary place to work through a decision end to end.

Where can I read more about how this works?

Start with SIGNLD extension for Claude for the full description of what the extension does and how it is built. Concepts explains how the Knowledge Graph itself is structured, and security covers the read-only access model, tenant isolation and inference controls in detail.

Try SIGNLD free or see how it works.

Claude and Anthropic are trademarks of Anthropic. Used descriptively only.

Key takeaways

  • The SIGNLD extension for Claude is built and maintained by SIGNLD using Anthropic's publicly available extension interfaces.
  • SIGNLD connects to business systems a company already runs, including spreadsheets, CRMs, ERPs and other operational tools, through read-only connections.
  • SIGNLD's own AI processing runs on a single-tenant AWS Bedrock instance dedicated to that tenant, under SOC 2 Type II controls, with tenant isolation enforced between customers, and it is never trained on a customer's data.
  • SIGNLD does not set or change Claude's pricing, usage limits or model behavior; those are Anthropic's terms and they change over time, so they are documented at docs.anthropic.com rather than restated here.
  • Start with SIGNLD extension for Claude for the full description of what the extension does and how it is built.