Getting started
Frequently asked questions
Common questions about the protocol. If yours isn't here, ask on X.
Is Pinecode a vector database I host?+
No. Pinecode is a protocol you query. There are no servers for you to run, no databases for you to provision. The protocol itself handles storage (IPFS), indexing (HNSW nodes), and settlement (Base + USDC).
How is this different from Pinecone, Weaviate, or Qdrant?+
Those are centralized vector databases. You pay them; they own your data; only you can search it. Pinecode is a public protocol where context is shared and monetized. Contributors are paid. Quality is curated by an open market.
Do I need to hold PINE to query?+
No. Recalls are paid in USDC via x402 micropayments. You only need PINE if you want to contribute new seeds (small bond required), curate by staking, or participate in governance.
What if the foundation goes away?+
The protocol keeps working. Contracts are on Base, content is on IPFS (3× pinned), and the indexer is open-source — anyone can run one. The foundation is a convenience, not a dependency.
Can I run my own indexer?+
Yes. The indexer is fully open source and permissionless. You earn a share of recall fees served through your indexer, plus you can sell premium SLAs (lower latency, larger limits) at your own price.
What stops someone from stealing my contributed seeds and re-listing them?+
Every seed has a content hash and a signature on-chain. Re-registering the same content fails on the contentHash uniqueness check. Republishing slightly modified versions can be challenged for duplication and slashed.
What's the privacy model?+
Pinecode is a public protocol — contributed content is public. For private memory you want a private indexer running locally that doesn't publish to the network. The same SDK supports both modes.
How does Pinecode handle copyrighted material?+
Contributors agree on submission to a content license (default CC-BY-4.0; license is part of the payload). DMCA-style takedowns route through governance, not through any individual operator.
What's the relationship with the Base / Coinbase team?+
Pinecode is independent. Base is the network we deploy on. We participate in Base's grants program and have received ecosystem support, but the protocol is not affiliated with or controlled by Coinbase.
I want to integrate Pinecode into my agent framework. Where do I start?+
If your framework already supports MCP (Claude, Cursor, most ai-sdk flows), just add the @pinecode/mcp server to your config. Otherwise, install @pinecode/sdk and call pinecode.recall() — it's one function.
How do you prevent low-quality AI-generated slop from drowning out good seeds?+
Three filters: (1) a bond per seed makes spam expensive; (2) recall-weighted quality scoring means useless seeds earn nothing; (3) curators with stake actively challenge bad content with slashing as enforcement.
Is the codebase open source?+
Yes. SDK, contracts, indexer, gateway, MCP server, and this docs site are all MIT-licensed and live at github.com/Pinecode-sdk/pinecode.
Docs v0.6.2