Sniparaproject brain
Menu
Product
Context layerProject context and memory for coding agents without a new chat surface.Project BrainDefinition, boundaries, and the project state an agent starts from.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

ChatGPT / OpenAI MCP + Snipara

Snipara connects to supported ChatGPT workspaces as a custom MCP app. It is not a Custom GPT Action and it does not expose a per-project OpenAPI schema.

Primary path
For LLM agents: use Hosted MCP
Use ChatGPT's Apps developer mode with the project-scoped Hosted MCP endpoint. Do not use an invented REST schema or project-health substitute.

Availability

Custom MCP app availability and write permissions depend on the ChatGPT plan and workspace policy. Workspace admins control developer mode, app review, action permissions, and publication. Verify the current controls in ChatGPT before starting; the labels may evolve while full MCP support remains in rollout.

Configure the custom app

Replace YOUR_PROJECT_SLUG with the value stored locally as SNIPARA_PROJECT_SLUG. A generated setup reference can fill the endpoint without putting credentials in source.

  1. Open ChatGPT on the web and enable Apps developer mode if your workspace permits it.
  2. In Settings or Workspace settings, open Apps and choose the option to create a custom app.
  3. Use the endpoint generated for your Snipara project:
    https://api.snipara.com/mcp/YOUR_PROJECT_SLUG
  4. Configure the authentication mechanism offered for your project. Keep the credential in the ChatGPT or workspace secret flow; never put a real key in a committed JSON file.
  5. Scan the tools, review the requested capabilities, create the draft, and publish it only after the workspace's approval checks pass.

Use and verify

  • Enable the Snipara app in a conversation before asking it to query project context.
  • Ask the app to call snipara_context_query for a source-backed answer, then verify the returned MCP content blocks rather than assuming the response is a plain JSON object.
  • For a structural code check, ask the app to use snipara_code_callers or snipara_code_imports when that specialist surface is available to the caller.
  • When the Hosted MCP contract changes, refresh or republish the app according to your workspace controls; approved tool snapshots do not necessarily update automatically.

Unsupported legacy paths

Do not configure Snipara as a Custom GPT Action. Health and contract verification belong to the supported Hosted MCP handshake and tool scan; there is no alternate per-project REST schema for this setup.

Related guides