synccontext logo

synccontext

by Flavio RobertoGitHubUpdated May 4, 2026

Shared team memory for AI coding agents. Save architecture decisions, patterns, and conventions — every team member's AI agent gets instant access via semantic search. 14 tools, multi-project isolation, pgvector, auto-detect embeddings (Gemini/OpenAI/Ollama).

mcp,
semantic-search,team-memory,pgvector,ai-agents,embeddings,claude-code
|

How to pay

Subscribe

Monthly billing

$19/month

Predictable monthly cost with included usage. Best for steady, high-volume traffic.

  • Unlimited tools within plan limits
  • One API key, billed once a month
  • Cancel any time
See plans

Overview

synccontext delivers a centralized repository for team-shared memories in development projects, enabling AI coding agents to access architecture decisions, coding patterns, bugs, and conventions via pgvector-powered semantic search. It isolates contexts by project, auto-generates embeddings from Gemini, OpenAI, or Ollama, and supports CRUD operations on memories tied to files, tags, authors, and types. This ensures AI agents and developers retrieve precise, relevant context instantly, reducing inconsistencies in multi-agent workflows.

Key Capabilities

  • save_memory: Stores content with optional author, tags (e.g., "auth", "frontend"), file_path, and memory_type (general, decision, bug, pattern, onboarding).
  • bulk_save_memories: Saves multiple memories in batch for efficient bulk uploads.
  • search_memories: Performs semantic search on query, with filters for tag, author, and top_k results (default 5).
  • find_similar: Retrieves top_k semantically similar memories to a given memory_id (default 5).
  • list_memories: Lists recent memories, filtered by limit (default 20), tag, author, or memory_type.
  • get_memory: Fetches a single memory by memory_id.
  • update_memory: Updates memory_id with new content (auto-re-embeds), tags, file_path, or memory_type.
  • delete_memory: Removes a memory by memory_id.
  • search_by_file: Finds memories matching a file_path substring (case-insensitive, limit default 20).
  • get_project_context: Returns a summary of the project's shared knowledge base.
  • list_tags: Lists unique tags with usage counts.
  • list_contributors: Lists contributors who saved memories.
  • list_projects: Lists all projects (requires admin_token).
  • create_project: Creates a project with name, optional description, and admin_token (returns token).

Use Cases

  1. A developer documents a React state management pattern using save_memory with tags ["frontend", "pattern"]; an AI agent later queries search_memories("state management best practices") to generate compliant code.
  2. During onboarding, a new team member calls get_project_context for an overview, then search_by_file("/src/api/user.js") to pull related API decisions and bugs.
  3. Refactoring a microservice: Use find_similar on a known database schema memory to locate matching patterns, then bulk_save_memories for new migration notes.
  4. Admin sets up isolation with create_project for a new repo, lists via list_projects, and monitors via list_tags or list_contributors.

Who This Is For

Development teams building with AI agents (e.g., Cursor, VS Code Copilot, or custom agents) on mid-to-large projects. Targets developers and engineering leads comfortable with API calls, UUIDs, and basic embeddings; no advanced ML expertise needed.

(298 words)