peepshow/ sinks/ typesense

Reel #80 Hybrid search

peepshow sink / typesense

TypesenseFast hybrid search — upsert runs into a Typesense collection.

Soft-creates the collection on first write and upserts one document per peepshow run keyed on `id == run_id`.

drop · process · typesense

What it does

[Typesense](https://typesense.org) is the open-source typo-tolerant search engine with hybrid (keyword + vector) search baked in. This sink soft-creates the collection on first write with a small schema (`run_id`, `title`, `frames`, `duration`, `transcript`, `thumbnail_url`, `strategy`, `created_at`), then upserts one document per peepshow run via `POST .../documents?action=upsert`. Re-running the same peepshow command replaces the existing row rather than 409ing. Pair with the Typesense dashboard or any of the official SDKs for sub-50ms typo-tolerant search across run history.

When to reach for it

  • Typo-tolerant search over peepshow run history — find runs by approximate transcript text or fuzzy title
  • Self-hosted or Typesense Cloud — one schema for hybrid keyword + vector search
  • Pair with Typesense's vector field on `transcript` for semantic recall via the same query

Install

npm i -g peepshow

Use it

TYPESENSE_URL="http://localhost:8108" \
TYPESENSE_API_KEY="xyz" \
peepshow ./demo.mp4 --sink typesense

Make it automatic

Register the sink once — every run fires it afterward. Scope by --when so it only runs for matching videos.

peepshow sinks add typesense
peepshow sinks add typesense --when extension=mp4,mov
peepshow sinks add typesense --when path=/Volumes/Work/

Configuration

  • TYPESENSE_URL Base URL of the Typesense node / cluster, e.g. `http://localhost:8108`. required
  • TYPESENSE_API_KEY Admin / write key. Sent in `X-TYPESENSE-API-KEY` header. required
  • TYPESENSE_COLLECTION Collection name. Default `peepshow_runs`. Auto-created on first write.
  • PEEPSHOW_FRAME_BASE_URL When set, the first frame URL is written to the `thumbnail_url` field.

Use with an LLM agent

Every peepshow sink reads its config from env vars and receives a single JSON payload on stdin. An LLM agent (Claude Code, Cursor, Windsurf, Gemini, Codex) can drive the Typesense sink automatically when three things are true:

  • the env vars below are exported in the agent's shell (or a project .env it can load),
  • the peepshow CLI is on PATH — install with npm i -g peepshow,
  • a peepshow auto-sink is registered for the run (optional but recommended — makes invocation zero-argument).

1. Set the environment

# Add to ~/.zshrc, ~/.bashrc, or a project .env the agent can load
export TYPESENSE_URL="..."
export TYPESENSE_API_KEY="..."

2. Register as an auto-sink

peepshow sinks add typesense
peepshow sinks add typesense --when extension=mp4,mov

3. Example LLM session

You → drop a .mov into Claude Code.

Claude → auto-invokes /peepshow:slides ./clip.mov. peepshow extracts frames + audio, the Typesense sink forwards the run to the configured collection. Claude replies with a summary and a link to the created record.

The transcript text is a natural embedding target — index it next to the frames.

Write your own

A sink is any executable that reads the --emit json payload on stdin. Shell, Node, Python, Go — the spec's in docs/PLUGINS.md. Register persistent ones with peepshow sinks add-cmd 'your-command'.