StackMap
Subscribe

rtk vs SoL-Pi

Rust CLI proxy compressing dev-command output 60-90% before your agent reads it — git, tests, linters, docker, 100+ commands; hooks auto-rewrite bash calls. Single binary, <10ms overhead. — versus — NVIDIA's Pi extension with four auto-research-discovered efficiency mechanisms - action fusion, observation packing, evidence-preserving log reduction, online context compaction - all opt-in.

The curated verdict

rtk shrinks dev-command output 60-90% at the shell layer for any agent; SoL-Pi's ObservationPack and Reducer tackle the same oversized-observation problem at the harness layer.

rtkSoL-Pi
Stars81k2.3k
Forks5.1k181
LanguageRustTypeScript
LicenseApache-2.0MIT
Last activity4 days agoyesterday
Topicscoding, localcoding, agents
Curated connections76

rtk — the curator's take

The rare token-saver whose README explains why its headline number is NOT your bill cut — bash output is one slice of input tokens, and the savings dilute; that honesty earns trust. Two things to know before judging your gains: the hook only intercepts Bash tool calls (Claude Code's built-in Read/Grep/Glob bypass it entirely), and `rtk gain` estimates tokens at bytes/4, not with a tokenizer. Compression is lossy by design — when a debugging session goes weird, the clue may be in the output rtk collapsed; `rtk proxy` is the escape hatch, use it. Beware the crates.io name collision with Rust Type Kit.

SoL-Pi — the curator's take

Install SoL-Pi if you run Pi as your coding agent and want fewer turns and less context replay without the agent skipping work: Action Fusion runs an edit's validation command in the same tool call, ObservationPack turns repeated big tool results into paged handles, the Reducer only compacts a log when every retained quote still matches the archive. It is an extension over unmodified Pi 0.85.1 - no patches, everything disabled until you write sol-pi.json. Not for Claude Code/Codex users (it uses Pi's extension APIs); headroom, rtk and token-optimizer are the harness-agnostic ways to get the same effect. The paper is the interesting part: the mechanisms were found by scaled auto-research loops over the harness itself, which is autoresearch/evo pointed at agent efficiency.