What it does

Context for the AI tools you already use.

AI assistants don't really understand your project. They fall back on grep and reading files one chunk at a time, and they have no idea your database schema changed last sprint. SocratiCode indexes code, dependency graphs and non-code knowledge into one context layer. On a 2.45M-line codebase, that approach burns 61% more context, takes 6–7× more tool calls, and runs up to 37× slower than a single semantic query against SocratiCode. Same AI, dramatically better answers.

1

Connect your repositories Sign in with GitHub, GitLab or Bitbucket and choose which repos to index, plus any schemas, API specs or docs you want in context. Zero configuration, zero local dependencies.
2

We index automatically: every push, every branch Webhooks trigger indexing on every push across every branch, so PR reviews see the code actually being reviewed. Non-code knowledge (database schemas, runbooks, API contracts) is indexed alongside source and stays in sync. Local changes are also indexed, so you also get uncommitted working-copy changes in the same search.
3

Your code stays yours Source is cloned ephemerally during indexing and deleted; only one-way embeddings and metadata persist. SSO, audit logs and data residency controls come built in, not as a later upsell.

GitHub

GitLab

Bitbucket

SocratiCode Context Engine

Hybrid search

Impact & call graphs

Cross-repo

Context artefacts

Secure & private endpoint

Claude Code

Cursor

Copilot

Windsurf

Any other public or private AI

Code Context Engine

The combined capabilities your assistant doesn't have on its own.

Individually, some of these exist in other tools. Together, in one consistent layer across every repo and every AI, they don't.

Hybrid search

Grep misses intent. Semantic search misses exact names. Hybrid fuses both via Reciprocal Rank Fusion into one ranked set, tested on codebases over 40 million lines.

Dependency & impact graphs

Your AI edited a function but didn't know three services call it. File-level and symbol-level call graphs across 18+ languages, with blast-radius analysis ("what breaks if I change X?"), call-flow tracing and circular dependency detection, mean it reasons about impact before it writes.

Cross-repo search

The bug is in the API gateway, but the AI only sees the front-end repo. Query every repository your organisation owns in a single call, so the assistant sees the full system.

Context artefacts

Your AI doesn't know the database schema changed last sprint. Index schemas, API specs, infra configs and architecture docs alongside code: knowledge that lives outside source, brought into context.

Branch-aware indexing

PR reviews that search stale main-branch data give stale answers. Every branch gets its own index, so reviews see results for the code actually being reviewed.

Web dashboard & visual explorer

Not everything belongs inside an AI chat. Search the index, explore an interactive graph of files and symbols with blast-radius overlay, browse artefacts, manage repos and team members. All visible and auditable in the browser, with PNG export for architects and security reviews.

Compliance-ready by default

Procurement won't approve a tool with no audit trail. SSO/SAML, audit logs, data residency controls, GDPR-ready DPA and subprocessor list from day one, not a premium add-on bolted on later.

Your code isn't training data

Other tools use your code to improve their models. We don't. Indexing is ephemeral, embeddings are one-way, and VPC or air-gapped deployments mean nothing leaves your network.

Open by design

Switch assistants tomorrow. Keep everything you've indexed.

Move from Cursor to Claude Code and lose every scrap of indexed context? That's the current reality. SocratiCode sits outside any single AI. Your search index, dependency graphs and context artefacts survive every tool switch, every model upgrade, every vendor change.

Every major AI assistant Claude Code, Cursor, VS Code Copilot, Windsurf, Codex and any other MCP-compatible tool, today and whatever ships next.
Private or local models Point it at your own LLM endpoint, including fully offline deployments on your own hardware.
One context layer, many assistants Half the team uses Cursor, the other half uses Claude Code? They all share the same indexed understanding of the codebase.

config

{
  "socraticode-cloud": {
    "url": "https://cloud.socraticode.com/your-org",
    "headers": {
      "Authorization": "Bearer sk-..."
    }
  }
}

Deployment

Hosted by us, or entirely by you.

Codebase security isn't one-size-fits-all. Choose the model that matches your risk profile. The underlying product is identical across all three.

Level 1

Managed cloud

You want zero infrastructure overhead. Just a URL.

Fully managed on AWS, in your choice of EU or US region. Multi-tenant or dedicated instance available.

• Ephemeral code cloning, then deleted
• AES-256 encryption at rest and in transit
• Data residency in EU or US
• Automated patching and maintenance

Popular with enterprise

Level 2

Your VPC

Your code can't leave your network, even temporarily.

Deployed into your own AWS, GCP or Azure account. Nothing ever leaves your network perimeter.

• Customer-managed encryption keys
• No Altaire subprocessors involved
• Your choice of embedding provider
• Deploys within your existing compliance perimeter

Level 3

Air-gapped on-prem

Your environment has no internet access at all.

For defence, healthcare and finance. Runs on bare metal, fully offline, with embeddings generated locally.

• Zero external API calls
• Behind your corporate firewall
• Fully localised LLM embeddings
• Complete data sovereignty

Governance

Ready for your security team.

The controls compliance teams expect from enterprise SaaS, built in from day one, not retrofitted later.

Centralised management

Every developer configuring their own AI tooling means no consistency and no oversight. One dashboard for repository access, team roles and billing puts you back in control.

SSO and SAML

A tool your security team can't provision or revoke through Okta won't get approved. Integrates with Okta, Azure AD and Google Workspace out of the box.

Audit logs

When compliance asks "who accessed what?", "we don't know" isn't an answer. Exportable logs of every query, every user and every repository access, available as an export or API feed.

IP allowlisting

Token auth alone doesn't satisfy every threat model. Restrict access to known networks on Enterprise plans for defence-in-depth on top of authentication.

Scoped access tokens

One shared API key with access to everything is a breach waiting to happen. Issue per-team or per-CI tokens with granular repository and capability scopes. Rotate or revoke in a click.

Compliance & frameworks

Enterprise procurement needs more than a promise. Security controls aligned with SOC 2 and ISO 27001 frameworks. GDPR-ready DPA and subprocessor list included from day one. Formal certification available on Enterprise plans.

Practical guides

Start with the decision, then the setup.

Plain-language guidance for engineering leaders, platform and security teams, followed by current setup paths for nine AI coding hosts.

Browse all guides Read the full FAQ →

Understand legacy code →

Recover context when documentation is missing: find the behaviour, trace dependencies, and check the impact of a change.

Shared team context →

What one shared index changes, what the open-source route requires, and when to discuss the managed beta.

Cross-repo context →

Why repository boundaries hide system behaviour, and what linked search can and cannot establish.

Source-code boundaries →

Separate indexing, vector storage, and assistant-provider paths before approving sensitive source.

Security review context →

Use search and graphs to navigate authorised source without treating them as vulnerability proof.

Design partners

Bring SocratiCode into the work you deliver.

We are exploring a partner-led route for consultancies and implementation teams that want to use or implement SocratiCode with their clients. This is a design-partner stage, not a finished reseller programme.

Discuss a design partnership Partner FAQ →

Delivery boundary

Define who owns discovery, implementation, integration, adoption, and the day-to-day client relationship.

Support and escalation

Define first-line support, product escalation, incident communication, and the evidence each side provides.

Contract and SLA

Define the supported service, measurable service levels, licensing, contracting, and commercial model before a standard partner offer is published.

Partner pricing, revenue allocation, contracting structure, and standard SLA commitments are not fixed. They are part of the design-partner work.

Direct Cloud beta plans

These plans remain the starting point for teams speaking to us directly about the Cloud beta.

Design-partner pricing, contracting, support ownership, and service levels are separate and are still being defined with partners.

Team

Stop every developer running their own Docker setup. One shared index, zero infra.

$39/mo base

+ $8 per user / mo

Includes

Shared team index
Cross-repo search
Branch-aware indexing
Up to 10 repositories

Apply for beta

Recommended

Business

Your procurement team needs audit trails and SSO before they'll sign off. This plan has both.

$99/mo base

+ $12 per user / mo

Includes

Everything in Team
Unlimited repositories
SSO / SAML
Audit logs
Usage analytics

Apply for beta

Enterprise

Your security policy requires code to stay on your infrastructure. We deploy there.

Custom

Annual billing

Includes

VPC or air-gapped deployment
Dedicated infrastructure
Security controls aligned with SOC 2 & ISO 27001 · DPA included
Bespoke integrations
Dedicated support engineer

Contact sales

Private beta: onboarding teams now

Apply for early access.

Leave your work email and we'll reach out with next steps. We're onboarding a small number of engineering teams during the private beta.

Thanks. Your request was sent. We'll be in touch.

Prefer email? Write to [email protected].

Open source at the core

An open-source engine, managed for your team.

The engine behind SocratiCode Cloud is open source under AGPL-3.0, audited by a public community and battle-tested across thousands of indexed repositories. That openness is your guarantee of transparency, security and engineering quality: every component that touches your code is inspectable, reviewed and proven in production before it reaches your team.

Cloud builds on the same engine, takes the operational work off your team and adds the controls compliance asks for:

Managed infrastructure (no Qdrant, Ollama or Docker to run, with automated patching and upgrades)
Webhook-driven auto-indexing on every push, every branch, every repo your organisation owns
Shared team context: one index for everyone, instead of a separate Docker setup per developer
SSO/SAML, audit logs, IP allowlisting and scoped tokens, with SOC 2 / ISO 27001-aligned controls, GDPR DPA, and EU or US data residency
VPC and air-gapped deployments, customer-managed encryption keys, bespoke integrations and a dedicated support engineer at the Enterprise tier

View on GitHub

Got questions or want to chat with other users? Join us on Discord →

Built-in AI search vs SocratiCode

Capability	Claude Code	Cursor	Copilot	+ SocratiCode
Text / grep search	✓	✓	✓	✓
Semantic search	—	✓	✓	✓
Hybrid search (fused)	—	—	—	✓
File-level dependency graph	—	—	partial	✓
Symbol-level call graph & impact analysis	—	—	—	✓
Interactive graph explorer	—	—	—	✓
Non-code knowledge	—	—	—	✓
Cross-repo search	—	—	—	✓
Shared team index	—	—	—	✓
Tool-independent (survives switching AI)	—	—	—	✓
On-prem / air-gapped option	✓	—	—	✓

Benchmarked on a 2.45M-line codebase in VS Code. Full methodology available on request.