Local MCP context · Available now

Stop letting AI agents guess your codebase.

Tessra gives Claude Code, Codex, Cursor, and other MCP clients local context from your real repo — symbols, dependencies, callers, and impact — before code changes.

7-day trial · 100% local · Full access Start trial → install Tessra → connect your agent → open a repo
Runs 100% local
Protocol MCP native
Best today Django · Angular · Flutter
CHAOS
STRUCTURED
MCP native · 11 clients supported

Works with the AI coding tools you already use

Tessra connects with your agent through MCP — no wrappers, no middleware.

Claude Code
VS Code
Cursor
Windsurf
Codex CLI/App
Gemini CLI
Kiro by Amazon
OpenCode
Also supports
RooCodeRooCode InsidersVS Code Insiders
Runs on
macOSWindowsLinux
Public benchmarks · Open source repos · 2026

Public benchmark results.

Four benchmarks on public, open source repositories — Angular, Django, and Flutter. Every prompt, evaluation criterion, result, and known limitation is available for review.

Budget model correctness
7%7%
Budget model without Tessra: 7% (1.75 / 25). With Tessra: 84% (21.00 / 25). ThingsBoard Angular · Claude Sonnet 4.6. Same prompt, same repo, same criteria. +77 pp.
Context beats blind exploration
budget + Tessra96%
>
premium alone53%
NetBox (Django): Haiku 4.5 + Tessra reached 96%. Sonnet 4.6 alone: 53%. Same task, same scoring.
Consistent across stacks
0/3stacks covered
AngularDjangoFlutter
Haiku 4.5·Sonnet 4.6
Angular, Django, and Flutter. Tested with Claude Haiku 4.5 and Sonnet 4.6. Tessra improved results on every evaluated stack.
Public benchmarks on open source repositories — ThingsBoard (Angular), NetBox (Django), NgRx platform (Angular/Nx), Ente Photos (Flutter). Prompts, evaluation criteria, results, and limitations available for review.
See the full public benchmarks →4 open source repos · Angular, Django, Flutter · Prompts, scoring, and results included
The problem

AI agents can read your files. They still miss the system.

Most coding agents still explore repo by repo, file by file. On real apps, that means wasted context, missed relationships, and risky changes.

01

They open too many files

Agents burn context exploring instead of reasoning.

02

They miss relationships

Routes, serializers, services, components, and tests rarely live alone.

03

Refactors become risky

The agent changes code before knowing what depends on it.

04

Onboarding stays slow

New devs and new agents both need a map of the repo.

How Tessra works

From repo to context in minutes.

Install Tessra, connect your MCP client, and let your agent ask for structured repo context instead of raw file dumps. Best today on Django, Angular, and Flutter.

STEP 01

Install Tessra

Copy your credential from the dashboard, install the CLI, and register the local service.

$ tessra install
STEP 02

Connect your agent

Run setup-ide to connect Tessra with Claude Code, Cursor, or any MCP-compatible client. Or add it manually via .mcp.json.

$ tessra setup-ide
STEP 03

Let your agent ask better questions

Claude Code, Codex, Cursor, or any MCP client receives structured context — symbols, callers, dependencies, impact — instead of raw file dumps.

Interactive demo

From question to safer change.

Ask your agent to analyze, refactor, or debug. Tessra gives it the context behind the answer: affected files, risky dependencies, and the path forward before code is edited.

tessra · decision engine
Agent message
CLAUDE-AGENT · 14:32
I need to refactor checkout discount logic without breaking webhooks.
Function in scope
Internal reasoning · abstract view
at risk affected related

This structure is computed internally — your agent receives only what matters.

Tessra · decision summary
ConfidenceHIGH
142 ms
7files affected
1webhook at risk
2test suites impacted
Suggested safer path

This change touches billing, checkout, and the Stripe webhookhandler. Refactor should preserve the existing webhook payload structure — downstream consumers depend on field order and types.

✓ contract preserved✓ tests updated! verify Stripe sandbox
checkout/views.py:84CALLER
api/orders.py:212CALLER
webhooks/stripe.py:47AT RISK
tests/test_discount.py:31TEST · 12
billing/refunds.py:62DEPENDS ON
Benefits

Less blind exploration. More useful context. Safer AI-assisted changes.

Tessra doesn't replace your agent — it gives it the structural context needed to reason before touching code.

Reduce blind exploration

Give your agent symbols, callers, dependencies, and impact — not another raw file dump.

Your code stays local

Tessra indexes locally. No cloud indexing. No MCP usage telemetry.

Framework-aware context for real apps

Full support for Django, Angular, and Flutter. More frameworks coming.

Django Angular Flutter + more
Security

Your code never leaves your machine.

Tessra runs locally. It indexes your repo on your machine and serves context to your MCP client without uploading your code or sending MCP usage telemetry.

  • Runs on your machine
  • No code uploads
  • No MCP usage telemetry
  • Works with your existing agents
100% local · MCP
Built for developers

Built for developers shipping AI-assisted changes in real repos.

FAQ

Common questions.

Does Tessra replace Claude, Codex, or Cursor?

No. Tessra is a local MCP context server. It gives your existing agent structured repository context — symbols, callers, impact radius — so it makes better decisions. You keep using the same agent.

Does my code leave my machine?

Never. Tessra indexes your repo and runs the MCP server entirely on your machine. No uploads, no cloud indexing, no external processing. The limit on the trial is enforced locally — Tessra doesn't send your code or MCP telemetry.

Does Tessra send telemetry?

Tessra does not send MCP usage telemetry — it does not track which tools your agent calls or what context it receives. Basic account events (registration, trial start, license checks) are handled server-side and documented in the privacy policy.

Why not just use grep or IDE search?

Grep and IDE search are useful when you already know what to look for. Tessra is for agent workflows where the model needs structured repo context: symbols, callers, dependencies, affected files, and impact before changing code. It complements your existing tools instead of replacing them.

What happens after the 7-day trial?

You can upgrade to Pro ($19/month) to keep full access. If you don't upgrade, Tessra stops serving context after the trial ends.

What's included in Pro?

Everything in the free trial — all MCP tools, local repo indexing, impact radius and call graph — plus email support and continued access after the trial ends. Pro is 1 user on 1 active machine, with multiple local projects.

Which projects does Tessra work with best?

Angular, Django, and Flutter — these three stacks have first-class support today with deep framework-aware indexing. Tessra works on other repos too, but results are best on these stacks.

Is Tessra a replacement for human code review?

No. Tessra surfaces impact, callers, and dependencies to help your agent make safer decisions. Final review and architectural judgment stay with you and your team.

Pricing

One local context layer. One simple price.

Full access for 7 days. No feature gates. Then $19/month to keep Tessra active.

Free Trial
$0 / 7 days

Full access to Tessra's core experience.

  • Full access for 7 days
  • Local repo indexing
  • All MCP tools
  • Django · Angular · Flutter
  • Value summary / local stats
  • No MCP usage telemetry
Start free trial

Tessra never sends your code or MCP telemetry. Everything runs locally.

Launch pricing · Limited spots
Pro
$19 / month

For developers using AI agents on real repositories every day.

  • One developer
  • One active machine
  • Multiple local projects
  • All MCP tools
  • Impact radius & call graph
  • Django · Angular · Flutter support
  • Email support
  • More stacks coming soon
Start Pro