Snipara Agents

Use Snipara to give AI agents durable repository context, reviewed memory, shared team state, and explicit task coordination. This is not a separate product wedge; it is the agent capability layer on top of Snipara repository continuity.

Product boundary

Agents use the same Snipara project, memory, MCP, and source-authority layer as the rest of the product. Hosted MCP and API remain the primary connection surfaces.

The Problem

AI agents today are stateless. Every conversation starts fresh. Past decisions, user preferences, learned patterns — all gone. This creates several challenges:

No memory persistence — Agents forget everything between sessions
No knowledge sharing— One agent's discovery stays siloed
No coordination— Multiple agents can't work together effectively
Repeated context loading — Same docs fetched over and over

The Solution

Snipara Agents is easiest to read as three blocks delivered via MCP (Model Context Protocol): durable memory, multi-agent coordination, and hierarchical tasks.

Agent Memory

Persistent semantic memory for individual agents with evidence links, verification, invalidation, and semantic recall.

Team Memory

Shared memory across agents and teammates. One agent's discovery becomes available to the team scope.

Multi-Agent Coordination

Real-time state sync, resource locking, and task handoffs. Build swarms that coordinate.

Hierarchical TasksNEW

Full project management with 4-level task hierarchies, policies, evidence requirements, and observability.

Enterprise OrchestratorADVANCED

Proof gates, drift checks, htasks, and explicit multi-agent coordination for Team Sync governance.

Snipara Sandbox

Sandboxed code execution, repeatable validation, optional agent execution, and trajectory logging.

How It Works

Snipara Agents exposes MCP tools that your LLM can call to store, retrieve, and share memory. The LLM decides when to remember something important and when to recall relevant context. Memory V2 also adds lifecycle controls so your agents can verify evidence, invalidate stale knowledge, and supersede old decisions safely.

// Agent remembers a user preference

snipara_remember({

  content: "User prefers TypeScript over JavaScript",

  type: "preference",

  confidence: 0.9

})

// Later, agent recalls relevant memories

snipara_recall({

  query: "What programming language does the user prefer?",

  limit: 5

})

Quick Start

Create a project

Get API key

Add Hosted MCP

Use agent tools

1. Create or open a Snipara project

Start from the dashboard or use create-snipara to connect a repository and configure Hosted MCP.

2. Get Your API Key

Navigate to your project's API Keys tab and create a new key. This key will be used to authenticate MCP requests.

3. Add MCP Configuration

Add Snipara to your MCP client configuration:

  "mcpServers": {

    "snipara": {

      "type": "http",

      "url": "https://api.snipara.com/mcp/YOUR_PROJECT_SLUG",

      "headers": {

        "X-API-Key": "YOUR_API_KEY"

4. Start Using Agent Tools

Once configured, your LLM will have access to memory and coordination tools. Ask it to "remember this for next time" or "recall what we discussed about authentication".

MCP Tools

Tool	Description	Plan
`snipara_remember`	Store a memory with type, category, and TTL	Pro+
`snipara_recall`	Semantic recall of relevant memories	Pro+
`snipara_memories`	List memories with filters	Pro+
`snipara_forget`	Delete specific memories	Pro+
`snipara_swarm_create`	Create a new agent swarm	Team+
`snipara_swarm_join`	Join an existing swarm	Team+
`snipara_claim`	Claim exclusive access to a resource	Team+
`snipara_release`	Release a resource claim	Team+
`snipara_state_get/set`	Read/write shared swarm state	Team+
`snipara_broadcast`	Send event to other agents	Team+
`snipara_task_create`	Create a task in the queue	Team+
`snipara_task_claim`	Claim a task from the queue	Team+
`snipara_task_complete`	Mark a task as complete	Team+
`snipara_htask_create`	Create hierarchical task (N0-N3)	Pro+
`snipara_htask_complete`	Complete task with evidence	Pro+
`snipara_htask_block/unblock`	Block/unblock with propagation	Pro+
`snipara_htask_metrics`	Throughput, aging, recovery metrics	Pro+

Highlighted rows are new Hierarchical Task tools. See all 17 htask tools →

Plans and availability

Agent memory, team memory, htasks, and coordination tools are plan-scoped capabilities inside Snipara. Use the pricing page for current plan packaging instead of treating Agents as a separate subscription.

Next Steps

Agent Memory

Learn how to store and recall memories for individual agents.

Team Memory

Share knowledge across multiple agents and team members.

Multi-Agent Coordination

Build agent swarms with resource locking and task queues.

Swarm Project ManagementNEW

Hierarchical tasks with policies, evidence, and observability.

Go to Dashboard

Configure project context and agent access.