Sniparaproject brain
Menu
Product
Context layerProject context and memory for coding agents without a new chat surface.Project BrainThe context layer your coding agent starts from.Before / afterWhat changes when the agent starts from project state instead of recall.60s demoWatch a cold ACME agent become a briefed agent before the first edit.Proof replaySee what changes when the agent starts from reviewed project state.InstallConnect the hosted MCP endpoint to the agent you already use.Agent surfacesHow the same Project Brain reaches the coding agents teams already use.
Developers
QuickstartConnect an AI coding agent in minutes.Project Brain modelWhat changed, why, impact, next step, safety.Hosted MCPTool contracts for agent clients.API ReferenceProgrammatic context and project intelligence access.GitHub AppRepository sync and PR Answer Packs.Multi-agentCoordinate swarms, claims, state, events and htasks.Snipara SandboxRepeatable execution when validation needs proof.Companion CLIWorkflow continuity for long agent work.
ProofPhilosophyPricingDocs
Sign inStart Free

Mistral + Snipara

Use Snipara Hosted MCP with Mistral Le Chat Custom MCP Connectors, Mistral Vibe, or LangChain JavaScript projects built on ChatMistralAI so Mistral starts from a Project Brain instead of a blank chat.

Primary path
Hosted MCP first
Mistral remains the model/client. Snipara provides project context, reviewed memory, and optimized retrieval through the hosted MCP endpoint.

Where Snipara fits

Step 1
Mistral Le Chat / Vibe / LangChain
Model client or app runtime
Step 2
Hosted MCP or bound Snipara tools
Project context interface
Step 3
Project Brain
Decisions, sources, and active work
Work Brief
Impact
Verification
Receipts

Snipara is the project intelligence layer. Mistral stays the model or client while Snipara supplies reviewed decisions, source-backed context, active work, impact, and verification receipts.

What changes inside Mistral?

Before

Implement OAuth.

The model relies on prompt context,
then asks or guesses its way
through the project.

After

Opening Work Brief...

OK Decisions
OK Active work
OK Architecture
OK Impact
OK Verification

Ready.

Why Mistral users install Snipara

  • Stop re-explaining the project.
  • Reuse reviewed decisions.
  • Know code impact before editing.
  • Start every session with a Work Brief.

Lifecycle integration

Mistral integrations use Hosted MCP, Vibe config, or LangChain-bound tools. Mistral request hooks are model request hooks, not local coding-agent lifecycle hooks; use Companion or an editor integration when you need local session, edit, commit, and verification lifecycle.

Generate the files

Run npx create-snipara --client mistral or choose Mistral in the interactive flow. The installer generates:

  • .vibe/config.toml
  • .snipara/templates/mistral-vibe-config.toml
  • .snipara/templates/mistral-le-chat-mcp.json
  • .snipara/templates/mistral-langchain-tools.ts
  • MISTRAL.md in the companion pack

Mistral Vibe

Merge the generated server block into .vibe/config.toml or ~/.vibe/config.toml, then restart Vibe.

[[mcp_servers]]
name = "snipara"
transport = "streamable-http"
url = "https://api.snipara.com/mcp/YOUR_PROJECT_SLUG"
api_key_env = "SNIPARA_API_KEY"
api_key_header = "X-API-Key"
api_key_format = "{token}"

Le Chat

In Le Chat, add a Custom MCP Connector with the Snipara hosted MCP URL for your project. Keep the key in your environment or secret store as SNIPARA_API_KEY.

LangChain ChatMistralAI

LangChain JavaScript supports Mistral tool calling with bindTools. The generated mistral-langchain-tools.ts file wraps snipara_recall, snipara_context_query, and snipara_settings as tools.

LangChain also exposes beforeRequestHooks, requestErrorHooks, and responseHooks on ChatMistralAI. These are model request hooks, not local agent lifecycle hooks like Claude or Codex hook scripts. Use snipara-companion recall, query, workflow, and task-commit when you need explicit local workflow fallback.

References