Skip to main content

​
Pinecone

Shanone’s Pinecone integration gives your agent 85 tools covering the full vector database surface — index and namespace management, vector CRUD/search, the Assistant API for RAG chat, backups/restore, and organization-level admin (users, service accounts, role bindings) most integrations skip.

​
Getting Started

1

Create a Pinecone API key

Generate a key from the Pinecone Console.
2

Add the key in Shanone

Go to Integrations in the Shanone dashboard, select Pinecone, and paste in the key (pcsk_...).
3

Retry your request

Once saved, shanone_execute_tool calls for pinecone_* tools will succeed.

​
Available Tools

Shanone provides 85 tools for Pinecone, organized into these categories:
pinecone_list_indexes, pinecone_create_index, pinecone_describe_index, pinecone_delete_index, pinecone_configure_index, pinecone_create_index_for_model, pinecone_list_namespaces, pinecone_create_namespace, pinecone_describe_namespace, pinecone_delete_namespace, pinecone_upsert_records, pinecone_search_records
pinecone_upsert_vectors, pinecone_query_vectors, pinecone_fetch_vectors, pinecone_delete_vectors, pinecone_update_vector, pinecone_list_vectors, pinecone_describe_index_stats, pinecone_fetch_vectors_by_metadata
pinecone_list_assistants, pinecone_create_assistant, pinecone_describe_assistant, pinecone_update_assistant, pinecone_delete_assistant, pinecone_chat_assistant, pinecone_chat_completion_assistant, pinecone_context_assistant, pinecone_upload_assistant_file, pinecone_list_assistant_files, pinecone_describe_assistant_file, pinecone_upsert_assistant_file, pinecone_delete_assistant_file, pinecone_list_assistant_operations, pinecone_describe_assistant_operation
pinecone_create_backup, pinecone_list_index_backups, pinecone_list_project_backups, pinecone_describe_backup, pinecone_delete_backup, pinecone_create_index_from_backup, pinecone_list_restore_jobs, pinecone_describe_restore_job
pinecone_start_import, pinecone_list_imports, pinecone_describe_import, pinecone_cancel_import, pinecone_list_models, pinecone_describe_model, pinecone_generate_embeddings, pinecone_rerank
pinecone_list_collections, pinecone_create_collection, pinecone_describe_collection, pinecone_delete_collection
pinecone_admin_list_organizations, pinecone_admin_get_organization, pinecone_admin_update_organization, pinecone_admin_delete_organization, pinecone_admin_list_projects, pinecone_admin_create_project, pinecone_admin_get_project, pinecone_admin_update_project, pinecone_admin_delete_project
pinecone_admin_list_users, pinecone_admin_get_user, pinecone_admin_delete_user, pinecone_admin_list_api_keys, pinecone_admin_create_api_key, pinecone_admin_get_api_key, pinecone_admin_update_api_key, pinecone_admin_delete_api_key, pinecone_admin_list_invites, pinecone_admin_create_invite, pinecone_admin_get_invite, pinecone_admin_delete_invite, pinecone_admin_resend_invite, pinecone_admin_list_role_bindings, pinecone_admin_create_role_binding, pinecone_admin_get_role_binding, pinecone_admin_delete_role_binding, pinecone_admin_list_service_accounts, pinecone_admin_create_service_account, pinecone_admin_get_service_account, pinecone_admin_update_service_account, pinecone_admin_delete_service_account, pinecone_admin_rotate_service_account_secret

​
Common Use Cases

Semantic search

Query an existing index for the nearest vectors to a piece of text or embedding

RAG assistant chat

Create a Pinecone Assistant over an uploaded knowledge base and chat against it directly

Index lifecycle management

Provision a new index for a workload, then back it up before a risky migration

Bulk ingestion

Kick off a bulk import from object storage instead of upserting vectors one batch at a time

​
Troubleshooting

The vector dimension must exactly match the index’s configured dimension — check pinecone_describe_index first, especially after switching embedding models.
These operate on Pinecone’s newer records API (namespace-scoped, with built-in embedding via integrated inference) rather than raw vectors — check shanone_get_tool_schema to confirm which field shapes each expects.
Confirm the file finished processing with pinecone_describe_assistant_file before chatting — newly uploaded files need to be indexed before the assistant can retrieve from them.
Admin endpoints require an API key scoped at the organization level, not a project-level key — generate a dedicated admin key from the Pinecone Console if you only have project keys configured.