95 typed Typesense operations
One typed operation registry powers a human-friendly CLI and a secure MCP server that is read-only by default.
I found 12 in-stock chairs under €500.
documents.search · 38ms · structured resultTwo interfaces. One source of truth.
Stop rebuilding one-off scripts and narrow integrations. Discover, validate, and run the same named operations from a shell or any compatible MCP client.
Profiles, readable tables, structured JSON, input schemas, examples, shell completion, and deliberate confirmations for destructive work.
Start with read-only search and inspection tools. Opt in to writes and administration only when the workflow calls for them.
One workflow, end to end
The demo is deterministic and credential-free. It mirrors the real operation names and output flow without connecting to your cluster.
$ tsk profile add production --url https://search.example.com
API key: •••••••••••• saved securely
$ tsk collections.retrieve --input '{"collection":"products"}'
18 fields · 24,891 documents · default_sorting_field: popularity
$ tsk documents.search --input '{"collection":"products","params":{"q":"oak chair","query_by":"title"}}'
✓ 12 matches in 38ms
agent Summarize the in-stock matches under €500.
TypesenseKit MCP · documents.search
I found 12 matching chairs. The strongest value options are:
Why TypesenseKit
Use the official client in application code. Reach for TypesenseKit when humans, automation, and agents need the same operational surface.
| Capability | curl scripts | Typesense client | Basic MCP wrapper | TypesenseKit |
|---|---|---|---|---|
| Terminal-first UX | Manual | — | — | Built in |
| MCP tools | — | — | Yes | Yes |
| Shared operation names | — | — | Varies | CLI + MCP |
| Input validation | You build it | Client types | Varies | Before requests |
| Safe defaults | You build them | Application-owned | Varies | Read-only + prompts |
| API escape hatch | Any endpoint | Client surface | Varies | api.call |
Real operational work
First-class operations target Typesense v30.2, with a raw API escape hatch for endpoints that land sooner.
List, create, update, delete, and wait for schema changes from one predictable interface.
Import, export, index, upsert, search, retrieve, and delete documents with structured inputs.
Run searches, multi-searches, facet exploration, suggestions, and saved preset workflows.
Operate aliases, presets, synonym sets, curation sets, stopwords, and stemming dictionaries.
Work with natural-language search models, conversation models, and conversation history.
Check health, metrics, stats, debug data, slow requests, snapshots, and maintenance operations.
Designed for operational safety
Typesense administration touches data and credentials. TypesenseKit makes the safe path the easy path for both humans and agents.
The MCP server exposes search, reads, collection metadata, and status checks unless full access is explicitly enabled.
Secure prompts, stdin input, macOS Keychain profiles, and redaction utilities reduce accidental API-key exposure.
The CLI confirms destructive operations. Non-interactive bypass requires a deliberate --yes.
Structured schemas catch malformed operation inputs before requests reach your Typesense cluster.
Open source · MIT
If TypesenseKit saves you a script, a dashboard detour, or an unsafe agent integration, star the repository so the next developer can find it.