feat: add Publication Pre-Flight Standard — the 8 P's

Present as precedent a performant, profiled, probed, proven, portable
program for production and publication.

7-section checklist: tests (P2P+E2E+aspect+build+exec+self-test),
static analysis (panic-attack+Hypatia), formal proofs (Idris2+Lean4+Agda),
bibliography quality (real authors, impact venues, recency), claims
verification (reproducible numbers, no overclaims), Zenodo/HAL metadata,
and final review (reproducibility, spellcheck, LaTeX).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: hyperpolymath

publication-pre-flight/PREFLIGHT.adoc

@@ -0,0 +1,181 @@
+// SPDX-License-Identifier: PMPL-1.0-or-later
+// Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>
+= Publication Pre-Flight Standard
+Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
+:toc: macro
+:sectnums:
+
+Every hyperpolymath paper MUST pass this checklist before submission to
+Zenodo, HAL, arXiv, or any journal/conference. No exceptions.
+
+[quote]
+____
+**Present as precedent a performant, profiled, probed, proven, portable program for production and publication.**
+____
+
+The 8 P's: **Present** (publishable), **Precedent** (novel contribution),
+**Performant** (benchmarked), **Profiled** (no pathological hotspots),
+**Probed** (panic-attack clean), **Proven** (machine-checked proofs),
+**Portable** (builds on CI, not just your machine),
+**Production** (ready to use, not a toy).
+
+toc::[]
+
+== 1. Test Suite (ALL must pass)
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| Point-to-point (P2P)
+| Individual function/module tests. Every public function has at least one test.
+
+| End-to-end (E2E)
+| Full pipeline tests. Input → output matches expected for representative cases.
+
+| Aspect tests
+| Cross-cutting concerns: security (injection, XSS, SSRF), performance (no O(n²) where O(n) claimed), concurrency (no data races), accessibility (where applicable).
+
+| Build tests
+| `cargo build`, `mix compile`, `zig build`, `idris2 --check` — zero errors, zero warnings treated as errors.
+
+| Execution tests
+| Binary runs, produces expected output, exits cleanly. CLI --help works.
+
+| Self-tests
+| If the tool has built-in diagnostics (`opsm check`, `panic-attack assail`), run them on the repo itself.
+
+| Benchmark suite
+| If the paper claims performance numbers, benchmarks MUST exist in repo and be runnable. Results must match claims within 20%.
+|===
+
+== 2. Static Analysis
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| panic-attack assail
+| Zero critical findings, zero high findings. Medium findings documented.
+
+| Hypatia scan
+| `.hypatia/` directory present with recent scan cache.
+
+| Dangerous patterns
+| `grep -r "believe_me\|sorry\|Admitted\|assert_total\|postulate\|unsafeCoerce\|Obj.magic"` — ZERO in proof code. Crypto axioms (SHA256 collision resistance etc.) are permitted IFF documented in paper with justification.
+|===
+
+== 3. Formal Proofs
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| Idris2
+| All `.idr` files compile with `%default total`. No `believe_me`, `assert_total`, or `postulate` except documented crypto axioms.
+
+| Lean4
+| If claims are core to the paper, cross-validate at least the key theorems in Lean4.
+
+| Agda / Coq
+| If the paper is in a domain where Agda or Coq is standard (e.g. PL metatheory), provide proofs there too. Zero `sorry` / `Admitted`.
+
+| Proof coverage
+| Every theorem stated in the paper must have a machine-checked proof in the repo. If a theorem is conjectured (not proven), it MUST be labelled "Conjecture" not "Theorem".
+|===
+
+== 4. Bibliography
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| .bib file exists
+| Co-located with the .tex file. Every `\cite{}` resolves.
+
+| Authors are real
+| Every cited author appears on DBLP, Google Scholar, or Semantic Scholar with a real publication history.
+
+| Papers exist
+| Every cited paper can be found via its DOI, URL, or title search. No fabricated references.
+
+| Impact quality
+| At least 50% of citations should be from venues with known impact (POPL, PLDI, ICFP, OOPSLA, LICS, TOPLAS, JFP, IEEE S&P, CCS, USENIX Security, or equivalent). Arxiv preprints are acceptable but should not dominate.
+
+| Self-citation
+| Self-citations are permitted but must be minority (<30%). Each self-citation must be genuinely relevant, not padding.
+
+| Recency
+| At least 30% of citations from the last 5 years. Classic foundational references exempt.
+|===
+
+== 5. Claims Verification
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| Quantitative claims
+| Every number in the abstract (N modules, X% improvement, Y tests) must be reproducible by running code in the repo. `wc -l`, `grep -c`, test output, benchmark logs.
+
+| No overclaims
+| "We prove X" means a machine-checked proof exists. "We show X" means experimental evidence exists. "We conjecture X" means neither. Never use "prove" for "argue" or "demonstrate".
+
+| Comparison claims
+| "Better than X" or "unlike X" claims must cite X and explain the specific dimension of comparison. No strawmen.
+
+| Novelty claims
+| "First to X" or "no existing system does X" — verify via thorough literature search. These are the claims most likely to be wrong.
+|===
+
+== 6. Zenodo / HAL Metadata
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| `.zenodo.json`
+| Present in repo root with title, description, creators, keywords, subjects, license.
+
+| HAL metadata
+| `hal-metadata.yml` present with title, authors, abstract, domains, keywords.
+
+| GitHub release
+| Tag `vX.Y.Z-paper` created. Zenodo auto-creates DOI from release.
+
+| DOI badge
+| After first Zenodo sync, add DOI badge to README.
+
+| Dual deposit
+| Submit to BOTH Zenodo (DOI) and HAL (French national archive) for maximum discoverability.
+|===
+
+== 7. Final Review
+
+[cols="1,3"]
+|===
+| Check | Description
+
+| Co-author review
+| If there are co-authors, all have reviewed and approved.
+
+| Licence headers
+| SPDX headers on all source files. Paper itself under CC-BY-4.0 for text, PMPL for code.
+
+| Reproducibility
+| A reviewer should be able to: clone the repo, run the tests, verify the claims, in under 30 minutes. If setup takes longer, add a `setup.sh`.
+
+| Spellcheck
+| Run `aspell` or equivalent on the .tex file.
+
+| LaTeX compiles
+| `pdflatex` + `bibtex` produces a clean PDF with no warnings.
+|===
+
+== Traffic Light Summary
+
+**GREEN** = all checks pass → submit immediately
+
+**YELLOW** = 1-3 fixable issues → fix in one session, then re-check
+
+**RED** = structural problems (missing implementation, unverified claims, wrong venue) → do not submit