🍯 Honey (I Shrunk the AI)

Write less code and say less about it. Honey (I Shrunk the AI) by GreenPT is a cross-tool coding skill that cuts AI coding-agent token usage and LLM API costs — making agents emit less code and less prose without losing correctness. It works with Claude (claude.ai and the API), Claude Code, Cursor, GitHub Copilot, Codex, Gemini CLI, Windsurf, Cline, OpenClaw, and Kiro. Three independent levers, applied reflexively:

1. Less code — YAGNI first. Walk a ladder (does it need to exist? → stdlib → language native → existing dependency → one line → minimum block) and stop at the first rung that works. The cheapest line is the one you never write.
2. Less prose — drop the wind-up, the hedging, the narration of code that already speaks for itself. Answer first.
3. Denser agent-to-agent handoffs — when the reader is another agent, not a human, hand it the most token-efficient format it parses losslessly (compact / columnar JSON, or ESON). Cuts handoff size ~in half at zero loss of recovery. Fires only here — never as a user-facing answer.

Honey combines what Ponytail (minimal code) and Caveman (terse prose) do separately, then goes further:

• Auto-intensity — lite / full / ultra chosen reflexively from the request, with no deliberation tax (it never spends reasoning tokens deciding how to comply — that would defeat the purpose on reasoning models).
• Safety carve-outs — input validation, error handling, auth, secrets, migrations, deletes, and anything you explicitly asked for are never compressed. Lazy ≠ broken.
• A skill family, not one prompt — an always-on core plus on-demand satellites (review, eco, gain, compress) and a hive of read-only subagents that return compressed handoffs. See Skills & subagents.

Why

Volume is cost. In agentic coding sessions, the volume of generated code and prose is what runs up the bill — and most of it is waste.

This repo ships a reproducible benchmark (bench/) so you don't have to take the numbers on faith: 23 tasks across three kinds of work — baseline vs Caveman vs Ponytail vs Honey — same model, same prompts, only the skill changes. Correctness is objective (unit tests, structural / accessibility checks, and lossless round-trip recovery for agent handoffs); quality is scored by a 4-model cross-family judge panel (median of Opus 4.8 + Sonnet 4.6

• Haiku 4.5 + GPT-5.5) under a neutral rubric that says nothing about length, so a terse skill gets no thumb on the scale. The figures below are the committed results (Claude Opus 4.8, 3 runs each) — run cd bench && npm run bench to reproduce.

A single blended number hides the story, because the levers fire differently per task type. Quality is % of baseline (panel median; for handoffs, lossless recovery); tokens are generated output vs baseline:

Honey leads quality where it matters most — it tops the user-facing and agent-to-agent tiers (the quality-separating ones) and stays within judge noise of the pack on saturated code tasks — while cutting tokens where it's safe to:

• Code — the deepest cut (−49% output) at essentially tied quality (98% vs 100%, within judge noise on tasks every variant passes). Caveman saves less; Ponytail's mandatory self-check inflates trivial code (+24%).
• User-facing — the carve-out keeps Honey from compressing polish, yet it still trims output (−6%) while earning the top quality score (101% of baseline) and the only 100% accessibility pass; Ponytail strips hardest and drops to 81% on the structural/a11y checklist.
• Agent-to-agent — under adversarial relay queries (ordinal, nested, absence, cross-field count) Honey is the only variant that stays 100% lossless while roughly halving handoff size (−51%); Caveman and Ponytail compress harder and lose recovery (67% / 50%). Its biggest, cleanest win.

The same pattern holds on GPT-5.5 (full two-provider table in bench/results/cross-provider.md): Honey is the only variant with no test regressions across all three tiers on Opus, and on both models it keeps top-tier quality while cutting tokens on every tier.

End-to-end agentic measurement (Cline harness)

npm run bench makes one API call per task — clean for isolating the output lever, but it never exercises an agent loop, tool schemas, or multi-turn context growth, where a real agent's token bill actually lives. bench/src/cline-bench.js (npm run bench:cline) runs each task through the Cline CLI headless, so the measured tokens are end-to-end agentic — harness prompt and every loop iteration included. Honey is injected as a Cline rule, recommended as the per-turn-cheap skills/honey/cline-rule.md (the operational core; the full SKILL.md re-sent every turn inflates input). See bench/README.md.

ESON — Efficient Structured Object Notation

Honey includes ESON, a zero-dependency, schema-first format for agent handoffs. Repeated record keys are emitted once; declared row counts catch truncated messages; JSON-compatible cells preserve types. ESON is developed in its own repo — Green-PT/honey-eson: the normative spec, JS + Python reference implementations, conformance vectors, the canonical LLM primer, the Honey Wire Profile, and negotiation. Honey vendors the codec in eso/.

The reproducible ESON/TOON/JSON benchmark measures bytes, two tokenizer estimates, codec speed, and lossless recovery across five agent handoff shapes. Run it with npm run bench:eso.

printf '%s' '{"from":"reviewer","findings":[{"sev":"H","issue":"expired token"}]}' | eson encode
eson decode < handoff.eson

CCR — for huge, redundant array tool output

ESON is lossless, for handoffs where every row matters. CCR (Compress-Cache-Retrieve) is the lossy-but-recoverable lever for the opposite case: a long uniform array you must read but mostly skim — logs, scan results, event streams. It keeps an informative sample (endpoints, anomalies/change-points, head/tail), caches the dropped rows locally, and leaves a <<ccr:HASH N_rows_offloaded>> sentinel. Nothing is lost — retrieve restores the original by hash on demand.

some-tool | eson crush          # → sampled view + sentinel; originals cached in .honey-ccr/
eson retrieve <hash>            # → the full original array, verbatim

Validated on a 90-row log (opus-4.8 + gpt-5.5): −82% tokens, crushed-only 96% answer accuracy, 100% with retrieve — and the lone crushed miss was a refusal, not a hallucination. Benches: npm run bench:ccr (tokens) and npm run bench:ccr:comprehension (quality). The honey-ccr skill tells the agent when to reach for it.

PX — image-rendered reads for huge dense read-only bulk

The intuition: sending a file as text pays per character; sending an image pays per pixel, no matter how much text is crammed into it. So a "photo of the page" costs ~5× less than the page itself — and reading it has photo problems: the gist survives, an exact serial number might not.

Concretely: dense text packs ~3 chars per image-token vs ~1 as text. PX exploits the gap on the read path: when the agent must skim something huge it will never edit (vendored code, a large diff, docs), it renders it to PNG pages with pxpipe's export and Reads the images instead of the text.

npx pxpipe-proxy export --json --out "$TMPDIR" src/   # → page-*.png + factsheet.txt + token report

Measured: up to −85% tokens on a single read. Repo-corpus bench (npm run bench:px, results): −79…85%, −82% average (26.4k Claude text tokens → 4.8k image est.); ~−75% all-in per read after the factsheet + report overhead; pxpipe's own end-to-end proxy bill measures −59…70% at whole-workload level.

Comprehension is a Fable story. The live 4-model panel (node bench/px/comprehension.mjs — 10 byte-exact questions, text vs render):

Only Fable-class models read renders usably — and even Fable is not byte-safe. Lossy on exact strings — misreads are silent confabulations (Haiku answered a seed question with 0x9e3779b9, a constant that isn't in the file), so the export ships verbatim precision tokens (paths, SHAs, numbers) as factsheet.txt text, and the honey-px skill forbids it for files you'll edit, secrets, or non-Fable readers. Over the raw API, prepend the export's prompt.txt banner — Fable's safety layer refuses naked dense renders. Complementary to CCR: CCR drops redundant rows recoverably; PX keeps everything in view at pixel prices. At /honey ultra the core skil