open-standards/jouleclaw
Repository files (latest commit first)
Filename Latest commit message Latest commit date
Transaction Science 4d7edf3b36 trx, screening: resolve XML entity references, which quick-xml 0.41 now reports as their own event
The quick-xml 0.41 bump did not just rename `BytesText::unescape`. 0.41 reports
every entity reference — `&`, `'`, `&` — as its own
`Event::GeneralRef` instead of folding it into the surrounding text. Both XML
readers here matched only `Event::Text`, so after the bump each reference was
dropped **silently**: the document still parsed, every existing test still
passed, and the values were wrong.

That is the dangerous shape of this change, and it is why it needed more than a
compile fix:

  - `op-screening` parses the OFAC/EU/UN sanctions lists. An entry published as
    "Smith & Co" was being stored as "Smith  Co" — a name that no longer matches
    the party it names, with nothing downstream able to tell. A screening miss
    reads exactly like a clean screen.
  - `jouleclaw-transform-trx` carries JSON property values through TRX XML. The
    quotes arrive escaped, so dropping the references turned `{"k":1}` into
    `{k:1}`, which fails to parse and falls back to being stored as a bare
    string. One round-trip test caught this; nothing else did.

Both readers now accumulate an element's character data across events and
resolve each reference: a numeric reference through `resolve_char_ref`, a named
one through `resolve_predefined_entity`. TRX carries no DTD, so an unknown name
there is refused rather than dropped. The sanctions readers keep an unknown
reference in its literal `&name;` form, so the anomaly stays visible in the
record instead of silently altering a name.

`trim_text(true)` had to go with it, on all four readers. Trimming is applied
per text event, so once a run is split by a reference the trim eats the spaces
on either side of it — "Smith & Co" became "Smith&Co", which the new tests
caught. The assembled value is trimmed once where it is consumed instead; the
screening readers already did exactly that at every consumer, making the
reader-level trim redundant there even before this change.

Tests: the fixtures had no entity in them anywhere, which is precisely why a
silent data-corruption bug could pass CI. Added, failing first: a sanctioned
name carrying `&` and one carrying `&` (different code paths — a named
predefine and a numeric reference), entity references in TRX text asserted
against hand-written XML rather than our own writer's escaping, and a refusal
for an unresolvable name.

Co-Authored-By: Claude Opus 5 <[email protected]>
2026-10-22 12:45:52 -04:00
..
conformance/v1 jouleclaw SPEC.md: normative §§8-12 for waves 12-21 2026-06-02 10:42:25 -04:00
jouleclaw-rs trx, screening: resolve XML entity references, which quick-xml 0.41 now reports as their own event 2026-10-22 12:45:52 -04:00
scales compliance: publish the EU draft PUE bands as a sourced data pack, not as code 2026-09-06 06:24:25 -04:00
vocab jouleclaw: wave-24 — contract + handoff + mission + value report 2026-06-03 02:35:34 -04:00
LICENSE jouleclaw: standard skeleton + Phase 1 foundation crates 2026-05-26 09:55:11 -04:00
LICENSE-SPEC licences: ship the Apache-2.0 text the code already declares 2026-10-09 17:22:32 -04:00
README.md licences: ship the Apache-2.0 text the code already declares 2026-10-09 17:22:32 -04:00
ROADMAP.md docs: strip vanity metrics from README + ROADMAP 2026-06-02 07:30:50 -04:00
SPEC.md energy meters: a GPU counter reading low is refuted in sandbox-meter, eoc-meter and jcp-energy too 2026-10-20 04:15:36 -04:00

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