exact-hold-pass
Skill4 starsChanged 21 days agoMITView on GitHub
Exact final-pass Toffoli reduction for reversible/quantum op streams: replace repeated unconditional Toffolis with a CX copy from a held wire (product holds), and clear-then-rebuild idle live wires whose value another live wire duplicates (dup loans, Clifford-only). Uses affine symbolic value tracking with a random 128-bit hash, a provable width cap, and fail-closed relocation. Use when a stream recomputes products or carries duplicated live values over idle intervals.
In this repository79 other files it gives its agents.
- AGENTS.mdAGENTS.md
- accepted-proof-exact-retuneSkill
- arxiv-portable-researchSkill
- auto-autoresearch-pipelineSkill
- beachhead-ratchetSkill
- bounded-commuting-foldSkill
- burn-the-house-downSkill
- calibrated-static-cost-proxySkill
- circuit-edit-integration-safetySkill
- colab-laneSkill
- computer-algebra-oraclesSkill
- constraint-construction-searchSkill
- cross-model-review-gateSkill
- deepseek-pressure-testSkill
- ecdsafail-fastevalSkill
- ecdsafail-lowq-playbookSkill
- ecdsafail-oracle-loopSkill
- ecdsafail-pareto-ratchetSkill
- ecdsafail-scratch-dietSkill
- elicit-researchSkill
- evaluator-scoring-surface-auditSkill
- exact-evaluator-replicaSkill
- exact-postpass-stackSkill
- exact-search-engineeringSkill
- fast-lean-proof-loopSkill
- fit-probe-and-spliceSkill
- gpu-affine-table-synthesisSkill
- greedyfrog-methodSkill
- hard-problem-contractSkill
- hold-demand-ledgerSkill
- hosted-carry-lendingSkill
- inert-op-screen-to-structural-ruleSkill
- invariant-chartSkill
- jev-second-perspectiveSkill
- lambda-exchange-deskSkill
- leader-rebase-stackSkill
- lean-arklib-proofsSkill
- lean-finite-certificate-splittingSkill
- lever-switches-and-break-controlsSkill
- mission-commandSkill
- nasqret-playbookSkill
- paged-walk-descentSkill
- peak-lifetime-ledgerSkill
- pinned-table-stackingSkill
- pip-disciplineSkill
- portable-source-assertionsSkill
- proximity-prize-campaignSkill
- proximity-prize-frontier-archaeologySkill
- quantum-affine-lookup-synthesisSkill
- quantum-borrowed-wire-counterexampleSkill
- quantum-dense-qromSkill
- quantum-dirty-borrowingSkill
- quantum-divstep-reconstructionSkill
- quantum-encoded-arithmeticSkill
- quantum-graded-divider-placementSkill
- quantum-guarded-permutation-auditSkill
- quantum-mbundo-lifetimesSkill
- quantum-record-miningSkill
- quantum-replay-arithmeticSkill
- quantum-residue-scan-ambiguitySkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 24
Product hold: If an unconditional CCX(a,b -> t) is recomputed later and the product still lives on a held copy, replace the repeat with a …
- Distinctive: line 26
Place a hold only if (live count + already-placed holds) ≤ width cap over the *whole* hold interval, and the copy sits on a wire provably …
- Very specific: line 28
Dup loan: If a live wire is idle over an interval and another live wire holds the same value (or its complement) at both interval ends, …
- Distinctive: line 41
Fail closed: if any relocation step cannot prove its precondition, the pass returns the stream unchanged, not a best-effort edit.
- Distinctive: line 46
Run it last, after every statistical pass, before the nonce tail — earlier passes change the stream and invalidate the intervals.
- Distinctive: line 48
Verify with a paired exact evaluation against the parent: expected signature is dT < 0 with dλ = 0.000 and byte-identical failing-shot sets.
- Distinctive: line 60
Do not confuse with measured-lifetime elimination at fixed width: their worker logged 38,665 such attempts with 0 committed.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Pfandleihi (exact-hold-pass)
Exact Hold Pass (product holds + dup loans)
Distilled from haotang1995's hold_pass (ecdsa.fail record submission e5f4dddc,
2026-10-29): 17,919 dup loans + 53 product holds, −53 T exact (dλ 0.000,
identical failing shots), implemented as the last pass before the nonce tail.
Reusable on any reversible op stream with repeated products or idle live wires.
The two primitives
- Product hold. If an unconditional
CCX(a,b -> t)is recomputed later and the product still lives on a held copy, replace the repeat with aCXfrom the copy. Place a hold only if (live count + already-placed holds) ≤ width cap over the whole hold interval, and the copy sits on a wire provably |0⟩ at placement. - Dup loan. If a live wire is idle over an interval and another live wire holds
the same value (or its complement) at both interval ends, clear the wire with
CXat the start and rebuild it withCXat the end. Clifford-only; frees width that product holds can spend.
Why it is exact — the correctness machinery
- Track every wire's value symbolically as an affine form over atoms: inputs, measurement outcomes, and products of affine forms.
- Fingerprint forms with a 128-bit random linear hash; equal hash ⇒ identical function up to collision probability ~2^-128. "Provably |0⟩" means form 0.
- No measurement is added and no RNG draw changes — the random-coin schedule stays aligned, so the nonce machinery upstream is untouched.
- Fail closed: if any relocation step cannot prove its precondition, the pass returns the stream unchanged, not a best-effort edit.
Operating rules
- Run it last, after every statistical pass, before the nonce tail — earlier passes change the stream and invalidate the intervals.
- Verify with a paired exact evaluation against the parent: expected signature is dT < 0 with dλ = 0.000 and byte-identical failing-shot sets. Any λ movement means the symbolic model leaked; revert and autopsy.
- The exactness caveat is the hash, nothing else: exact up to ~2^-128. Say so in the note.
Where to hunt for holds
- Repeated unconditional products emitted by template bodies (compute/inverse pairs that re-derive the same intermediate).
- Long idle intervals on live wires near the width peak — a dup loan there converts dead width into hold capacity exactly where holds need it.
- Do not confuse with measured-lifetime elimination at fixed width: their worker logged 38,665 such attempts with 0 committed. Loans and holds pay because they are value-equality driven, not liveness driven.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

