@typesensekit/cli

An open-source Typesense CLI for collections, documents, schemas, search, configuration, and cluster administration.

Typesense CLI overview · Complete CLI guide · GitHub

pnpm add -g @typesensekit/cli
# Secure interactive prompt
tsk profile add local --url http://localhost:8108

# Non-interactive stdin (avoids argv and shell history)
printf '%s' "$TYPESENSE_API_KEY" | tsk profile add ci --url https://typesense.example.com --api-key-stdin

# macOS Keychain-backed profile
tsk profile add production --url https://typesense.example.com --keychain

tsk profile test production
tsk profile export production > profile.json       # omits plaintext keys
tsk profile export production --reveal > backup.json
tsk profile import backup.json

Operation results use readable tables where possible. Pass --json for stable, redacted JSON output in scripts.

Destructive operations prompt before execution. Pass --yes only for deliberate non-interactive automation.

Every operation supports --schema and --examples. Enable shell completion with:

source <(tsk completion zsh)              # zsh
source <(tsk completion bash)             # bash
tsk completion fish | source              # fish

Discover operations and generate integration snippets:

tsk collections.list --input '{}'
tsk skills mcp

See the API coverage inventory for supported Typesense operations.

Creating API keys

keys.create redacts the new key by default. To retrieve it explicitly from the CLI, use --reveal and store the output securely:

tsk keys.create --input '{"value":{"description":"Search only","actions":["documents:search"],"collections":["products"]}}' --reveal --json

MCP output remains redacted. New or updated Keychain-backed profiles use unique credential references, so profiles with the same name in separate config files do not overwrite each other's keys. Existing references remain readable; saving a legacy profile again with --keychain migrates it to a new reference.