pi-spark

Measures how many prompt tokens pi's built-in system prompt and tools cost, and how many a custom tool or extension adds. Use when estimating or budgeting token usage for pi sessions, tools, or extensions.

Measures model prompt-cache hits, latency, and response-chain correctness across session topologies. Use when testing whether same-session, cross-session, or background model calls such as title and recap affect a main agent thread's cache behavior.

Releases pi-spark to npm via Release Please. Use when asked to publish a new version or apply a patch/minor/major or explicit-version release.