pinned-table-stacking
Skill4 starsChanged 21 days agoMITView on GitHub
Author declarative circuit-edit tables that are pinned to an exact gate list by length and hash, applied by an applicator that refuses any other stream, frozen and qualified as individual rungs, and stacked in call order before the nonce tail. Includes the portfolio-vs-exact labeling test (identical wrong-output and aligned failing-shot sets) and the "tables as durable assets" rule: a table above today's strict gate ships unchanged later inside a bigger stack. From jackylee0424's spookyfrog series on ecdsa.fail.
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-hold-passSkill
- 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
- 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
Files
Tällerturm (pinned-table-stacking)
Distilled from jackylee0424's public ecdsa.fail notes (spookyfrog 22/23, ba8ad256 / 2120d8a2, Oct 2026): his rows are tiny, safe, and constant — because the safety lives in the table format, not in reviewer heroics.
The table format is the safety mechanism
- Each table is a separate declarative artifact (his:
affine_followup.txt,union_followup.txt) whose header pins the exact gate list it applies to: the operation count and hash of the stream it expects. - The applicator (
source_edits) refuses to run on any other gate list. A table can never silently mis-apply after a parent change; it either matches the pinned stream or fails closed. - Tables apply in call order from the build file, after every inherited table and before the unchanged nonce tail. Inherited tables are kept byte for byte — you only ever add later tables.
Rungs, not commits
- Build each table on the current tree, then freeze it, qualify it, and get it independently reviewed as a rung of its own before it joins a stack.
- A rung whose price is above the board's strict gate is not deleted and not resubmitted alone — it ships unchanged later inside a bigger stack that clears the gate. Tables are durable assets that wait for a stack to carry them. (spookyfrog 23's first table was built on spookyfrog 22, re-bound after an inline table landed, and shipped a cycle later, unchanged.)
Portfolio vs exact — label honestly
- Exact means: identical wrong-output sets and aligned failing-shot sets against the parent on the tested inputs, with no measurement added or removed (his test: 336 inputs in both frames; first-table rows on 24 packets).
- A table that trades anything statistically is labeled portfolio, not exact,
and the note says so in the first paragraph. The paired failing-shot deltas ship
with uncertainties (
+0.020 ± 0.081), not bare zeros.
Devices worth farming as rows
Live-product rows, two-wire parity rows, held ANDs, and swap holds with relays — including holding a two-AND square term in place of a single-AND hold it displaces. Allocate each device jointly with the table's slot over storage shown clean without trusting any inherited sampled table.
Scorekeeping honesty
Report the board score immediately before submission and the exact improvement (his: −19,744, 0.002%), the ground nonce, the failing-shot rate with its paired comparison to the parent, and the Toffoli gate with headroom. Tiny accepted increments at a strict gate compound: his October run is ~1M score/day.
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.

