| Filename | Latest commit message | Latest commit date |
|---|---|---|
130123f648
All checks were successful
jouleclaw public surface must stay documented / Public-surface audit (jouleclaw-rs) (pull_request) Successful in 59s
standards workspaces build + test / which workspaces this change touches (pull_request) Successful in 33s
repository checks / lockfiles, tests in CI, workflow hygiene, spec and section references, public text (pull_request) Successful in 3m15s
standards workspaces build + test / macOS energy backends (pull_request) Successful in 10m2s
standards workspaces build + test / jouledesk-rs (jouledesk/jouledesk-rs, jouledesk-rs, macos) (pull_request) Successful in 3m49s
standards workspaces build + test / mesh-rs (mesh/mesh-rs, -p mesh-engine-gguf -p mesh-openai --features mesh-engine-gguf/gguf,mesh-openai/server, mesh-rs, linux-arm64) (pull_request) Successful in 3m3s
standards workspaces build + test / jouleclaw-rs (-nodebug, jouleclaw/jouleclaw-rs, -p jouleclaw-energy --features rapl,nvml, jouleclaw-rs, linux-arm64, -C target-feature=+fp16) (pull_request) Successful in 7m35s
Twenty-five source files in this crate were named by no `mod` line, so they were not in the build at all: ~19,000 lines that nothing type-checked, tested or linted. `scripts/check-orphan-sources.py` pinned them as a baseline so the count could only fall, and this is that fall: 25 -> 1. Declaring them costs almost nothing, which is the surprise. Twenty-three compile clean, and the two that do not are blocked by facts outside the code: video_decode.rs wants `ffmpeg_next`, which is not a dependency (`extern crate ffmpeg_next` -> E0463). That is a system library and a licensing choice, so it stays undeclared, and the baseline now holds it alone, with the reason written beside it. fusion.rs has every one of its `use` lines behind `#[cfg(feature = "metal")]` while the items that need those types do not, so a default build loses the imports and keeps the code: 31 unresolved types. It only ever compiled with that feature on. The module is now gated the same way, as `gpu_ops` and the sana_wm draft already are. sana_wm_gpu_dispatchers_draft.rs is gated likewise: it imports `crate::hal::metal`, which is `#[cfg(feature = "metal")]`. Measured on the build host, with --keep-going so an early error cannot hide what is behind it: `cargo check` 0 errors, `cargo clippy --all-targets` 0 errors, 44 tests pass, the whole jouleclaw workspace still checks, and the orphan guard passes on the new baseline, its planted-orphan self-test included. ⚠️ Separately, and not from this change: `--features metal` does not compile on main at all -- 3063 errors, concentrated in files that were already declared (hunyuan3d.rs 300, audiogen.rs 269, trellis.rs 213). Measured on origin/main before this change, so it is pre-existing. No CI job builds that feature. Co-Authored-By: Claude Opus 5 <[email protected]> |
||
| .. | ||
| conformance/v1 | ||
| jouleclaw-rs | ||
| scales | ||
| vocab | ||
| LICENSE | ||
| LICENSE-SPEC | ||
| README.md | ||
| ROADMAP.md | ||
| SPEC.md | ||
JouleClaw — Energy-Optimized AI Runtime
The substrate for AI that costs Joules, not tokens.
Capability per Joule, not capability per parameter.
What JouleClaw is
JouleClaw is a pure-Rust, omni-modal AI runtime + harness + agentic layer governed by Transaction Science as an open standard (5th in the open-standards quartet, alongside OpenPay, Smart Byte, EOC, and WAI).
Every operation JouleClaw performs walks an explicit energy-tiered cascade, top to bottom, stopping at the first stage that closes the query:
| Tier | Name | Cost class | What runs here |
|---|---|---|---|
| L0 | Cache | picojoules | Content-addressed cache hits |
| L1 | Lawful | nanojoules | Deterministic, pre-compiled primitive (text/code only) |
| L2 | Embed | sub-millijoules | Matryoshka embeddings + nearest-neighbor + hybrid search |
| L3 | Model | joules | Local SSM / ternary / multimodal / diffusion / etc. |
| L4 | Wire | tens of joules | Remote frontier RPC (escape hatch only) |
L1 (Lawful) is only meaningful for text and code modalities. For image, audio, video and 3D generation, the cascade collapses to L0 / L2 / L3 / L4 — the energy savings come from caching plus dispatch to the cheapest model that meets the quality bar.
Why JouleClaw exists
The mainstream "agentic AI" ecosystem treats inference as the entry point. Every state shift, every tool call, every routing decision runs through a 70-billion-parameter stochastic engine, burning kilojoules to do work that deterministic compute or a hot cache could resolve for microjoules.
JouleClaw inverts the default. Inference is the last resort, not the first. Frozen model weights are less trustworthy than current world state; fresh, provenance-stamped retrieval beats synthesis from training cuts. Every operation is metered in microjoules where the hardware permits it, and in millijoules where it doesn't — the spec is honest about the floor on every platform.
The thermodynamic circuit breaker is the safety mechanism. Set an energy budget; when measured consumption exceeds the budget, the breaker trips and the operation halts. Hallucination loops cannot survive an honest joule ledger.
What's here
| Path | Contents |
|---|---|
SPEC.md |
JouleClaw v1 spec — cascade tiers, energy contract, receipt |
CHARTER.md |
Stewardship pattern (shared across open-standards) |
jouleclaw-rs/ |
Apache-2.0 Rust reference implementation |
spec/ |
Tier specifications, energy provenance, receipt format |
conformance/v1/ |
Signed wire vectors for cross-implementation verification |
Reference implementation
jouleclaw-rs/ is the Apache-2.0 reference implementation in Rust. It
runs on Apple Silicon (M3/M4/M5 with UMA + Metal 4 Tensor APIs), AMD Strix
Halo / Ryzen AI Max, Intel Lunar Lake, NVIDIA Jetson, and discrete
NVIDIA / AMD GPUs.
It is omni-modal: text, code, vision, audio, video, image diffusion, 3D Gaussian splatting, and cross-modal fusion all dispatch through the same energy-tiered cascade.
The MCP tool surface is a JCP consumer: jouleclaw-jcp wraps any
jouleclaw-mcp metered tool so every dispatch is decided by the JCP runtime
against a signed, metered Grant before a byte reaches the tool — the tool's
identity is the content-addressed resource, the call's energy estimate is
debited from the grant's joule budget, and every decision (allow and deny
alike) seals as a JCR-1 receipt. An injected instruction can name an ungranted
tool; it cannot make the gate dispatch it.
Status
v0.1.0 reference implementation in active development. The spec is draft status. Cross-implementation conformance vectors will land with v1.0.0.
License
- Code: Apache-2.0 (see
LICENSE) - Spec text: CC-BY-4.0 (see
LICENSE-SPEC)