Smaller context. Byte-exact recovery.
Agents resend instructions, history and tool results on every turn. The Skill takes filler out of what the model writes; the local Proxy compresses what gets sent back and keeps the originals.
curl -fsSL https://raw.githubusercontent.com/JuliusBrussee/caveman/main/install.sh | bash- the skill
- 65% average output-token reduction across 10 prompts, in a 22–87% range
- scope
- output tokens only — input and reasoning are unchanged
- the proxy
- runs locally, compresses context, keeps the original bytes
- recovery
- byte-exact, from the local store
- fail-closed
- no parse, no recovery or no smaller output — it passes through
Local token reductions are estimates labeled inferred, never verified savings. No Caveman account is needed to run either.
Most of your context spend is rereading. The same pile, every turn.
It is the system prompt, the instructions, the whole conversation so far, and every tool result collected.
Log tails, JSON payloads, diffs and search results — the part nobody wrote, and the part that grows.
The model rereads the pile on every turn, to answer the one new line at the end of it.
Two places to cut. Neither loses the bytes.
- the skill, on output
- Removes filler from replies: 65% average output-token reduction across 10 prompts, in a 22–87% range. Output only — input and reasoning tokens are unchanged, and the skill itself adds prompt overhead.
- the proxy, on context
- Wraps agents you already run and sends their traffic through the local Caveman Engine, with matching compressors for logs, JSON, tables, code, diffs and terminal output.
- byte-exact recovery
- Original bytes are stored locally before lossy replacement, so the agent can retrieve exact detail later instead of guessing at what was there.
- fail-closed
- Parse failure, missing recovery or non-smaller output passes through unchanged. Nothing is compressed that cannot be given back.
What gets resent, and what gets a reference.
The one measured figure — 65% average output-token reduction across 10 prompts, in a 22–87% range — covers output tokens only. Local proxy reductions are estimates labeled inferred.
What powers it.
Send less.
Lose nothing.
curl -fsSL https://raw.githubusercontent.com/JuliusBrussee/caveman/main/install.sh | bashThe 65% figure covers output tokens only, across 10 prompts, in a 22–87% range. Local token reductions from the proxy are estimates labeled inferred, never verified savings, and local compression needs no Caveman account.