Skip to content

pplx-cli

Perplexity conversation archive & interactive query toolkit (pplx-export / pplx-ask).

It talks directly to the Perplexity REST/GraphQL API using your browser's logged-in cookies, and archives conversations — steps, citations, deep-research reports, computer-mode assets, and sub-agent workflows — as local Markdown + JSON. Raw API responses are always persisted first, so rendering can be re-run offline at any time.

Project purpose

Beyond archiving historical conversations, this project has a second goal: to bring local agents — the ones sitting closest to the data — a measure of Perplexity Computer's capabilities, so they can tune key parameters in code with high-quality information. (Perplexity ships no official Linux CLI, and this also puts a Max subscription to fuller use.)

It is not a full replacement for Computer mode. Two abilities remain out of reach:

  • the Deep Research skill's freedom to pick an arbitrary model;
  • the Council skill's ability to run deep research with several user-specified models, produce reports, and compare them side by side.

Features

pplx-export — archive your library

  • Library index and space index
  • Single-thread / batch export with incremental early-stop + resumable checkpoints
  • Asset backfill and credit-usage backfill
  • Conversation relation graph
  • Offline re-render (re-render, zero network)
  • Periodic-incremental cron snippet

pplx-ask — query Perplexity from the shell

  • SSE streaming queries in four modes: search / deep-research / council / study
  • Auto-move into a BOT space on completion, read receipts
  • Automatic archiving of every thread it creates — built so other agents can call it to fetch real-time information

Per-mode artifact boundaries (citations / reports / assets / sub-agents): see Modes; rendering-fidelity principles: see Export pipeline.

Where to next