Skip to content

docs(dag): defensive publication — ordered independent hard gates, non-compensatory evaluation, critical-failure cap - #268

Merged
harishquantamix merged 3 commits into
masterfrom
dag/w0-1-defensive-publication-public
Oct 3, 2026
Merged

harishquantamix merged 3 commits into
masterfrom
dag/w0-1-defensive-publication-public

Conversation

@quantamixsol

Copy link
Copy Markdown
Owner

What this adds

docs/dag/defensive-publication.md: a defensive publication (design disclosure) titled Ordered Independent Hard Gates with Non-Compensatory Evaluation and a Critical-Failure Cap for Governing Autonomous Agent Actions.

This is a documentation-only change: one new file, no code, no configuration, no version bump. The document describes the design symbolically: gate predicates and threshold names, with no configured values. Its merge commit is the public prior-art timestamp. The same text will be submitted to arXiv (cs.AI) on the day this PR merges, and the LaTeX source will cite the merge commit hash.

Text-unchanged check

The two commits are cherry-picked unchanged from the reviewed source. The check below proves three things:

  • (A) The file on this branch is byte-identical to the reviewed text.
  • (B) Its prose is identical to the pandoc source used for arXiv.
  • (C) Regenerating the LaTeX from that source reproduces the committed .tex with no differences, so the Markdown and the LaTeX carry the same text.

The arXiv source lives outside this repository.

[A] blob at HEAD          = dd55f0d3627952bbac20ee009d182c79f8aa86c9
    approved blob          = dd55f0d3627952bbac20ee009d182c79f8aa86c9
    A PASS: byte-identical
[B] Markdown prose (from '## 0.' to end) vs body.md prose: 310 vs 310 lines
    B PASS: identical
[C] pandoc(body.md + header.tex) vs committed defensive-publication.tex
    C PASS: identical
pandoc: pandoc 3.9

The only intended difference between the Markdown and the LaTeX is the document header. The Markdown header lines are replaced by a pandoc metadata block (title, author, date, abstract). The LaTeX adds one line, "First public disclosure: …", whose \disclosurecommit placeholder gets this PR's merge commit before submission.

$ git diff --stat origin/master..dag/w0-1-defensive-publication-public
 docs/dag/defensive-publication.md | 321 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 321 insertions(+)

Review

  • First approval (text review): done before this PR was opened
  • Second approval, Research Team: requested on this PR
  • Owner merges (this PR is never self-merged)
  • After merge: fill \disclosurecommit with the merge commit and date, then the owner submits to arXiv the same day

Governance

Step Result
Session gate opened (graq_lifecycle), graph 76,298 nodes
Safety check graq_safety_check(docs/dag/defensive-publication.md, add): risk medium (docs only), no safety boundaries hit
Scope one file; cherry-pick of the two reviewed commits, no edits
Push checklist git remote -v checked; push to origin on explicit owner authorisation for this branch only
Merge owner only

🤖 Generated with Claude Code

quantamixsol and others added 2 commits September 30, 2026 18:08
…s, non-compensatory evaluation, critical-failure cap

Design disclosure placing the DAG-2026 multi-gate design on the public
record as prior art. Symbolic only: HG-01..HG-07 predicates from CR-013
§2.3 with threshold names, never values. States plainly that no gate,
registry, cap or typology code exists; only the verdict schema (0.84.1+)
is public, with an empty reason-code seed and the flag off.

Private-first. Not for public posting until Research approves and the
owner OKs the GitHub post and arXiv submission.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 3266ad8470966bbc3fb4aa96776b6084384a5e03)
…rmed)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 8e65b4f1732b002b868b35628e75bf00931dc2ef)
@quantamixsol

Copy link
Copy Markdown
Owner Author

Research Team: please give the second approval on this defensive publication. It is one documentation file, byte-identical to the text you reviewed (see the text-unchanged check in the PR description: blob dd55f0d3, Markdown ≡ arXiv pandoc source ≡ regenerated LaTeX). Once you approve, the owner merges and the arXiv submission follows the same day.

@github-actions

github-actions Bot commented Sep 30, 2026 •

Copy link
Copy Markdown

🛡️ GraQle PR Guardian

💥 Blast Radius: 1 module affected

Module Files Changed Risk Level Impact Radius
docs 1 🟡 T2 1

Total blast radius: 1


🏛️ Governance Verdict

⚠️ WARN

  • T2: Advisory warnings detected. Review recommended.

🔍 SHACL Violations

No SHACL violations detected. ✅


🔐 Approval Requirements

This PR requires approval from: T2

  • A Senior Engineer or above must approve.

❌ Approval requirement NOT yet satisfied.


Metric Value
Blast Radius 1
Files Analyzed 1
Blocked 0
SHACL Violations 0
Verdict WARN

🔬 Powered by GraQle PR Guardian v0.1.0 · Scan completed 2026-10-03T08:22:46.874947+00:00

@quantamixsol

Copy link
Copy Markdown
Owner Author

Research Team Review: W0.1 defensive publication (second approval)

Reviewer: Research Team (GraQle Senior + Claude Code Junior)
Verdict: APPROVE
Reviewed: 2026-10-01T09:45:00Z, head 7e4b7a05

Text-unchanged check, reproduced independently

Check Result
Files in the diff exactly one: docs/dag/defensive-publication.md
(A) blob on dag/w0-1-defensive-publication-public dd55f0d3627952bbac20ee009d182c79f8aa86c9
(A) blob of the text approved in the first review dd55f0d3627952bbac20ee009d182c79f8aa86c9: identical
(B), (C) Markdown vs pandoc source, LaTeX regeneration not reproduced here (the arXiv source is outside the repo); accepted on the PR body's evidence

Every factual statement in §8 checked against its artefact

Statement Checked
tag v0.84.1 → 52e61661 yes
tag v0.85.0 → b861a7ed yes
0.85.0 is the current PyPI release (https://pypi.org/project/graqle/) yes, on 2026-10-01
the public reason-code seed is empty yes, _SEED = () at v0.85.0

§2 is dated "as of the drafting date", so it stays correct after 0.86.0 ships.

Trade-secret review: SAFE

The text contains threshold and cap names, the ranges (0, 1) and (0, 1], identifiers and dates, and no configured value. §7 lists what is withheld. It discloses no estimator for materiality or poisoning severity, and gives no reason for the gate order or the typology assignment.

What this unblocks

The merge commit becomes the first public statement of the multi-gate design. After that come the \disclosurecommit fill, the same-day arXiv upload, release 0.86.0 and lifting the RESEARCH.md / README §15 hold.

Recommendation

Ready for the owner to merge. Nobody self-merges.

Research Team · Senior consulted (0.826 tool / 91% synthesised, all agents in agreement)

…ling 2026-10-03; keeps the IP Content Gate intact)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit b5b04ef4f94b83bbba7d6008ce36ed860cd7c9b3)
@quantamixsol

Copy link
Copy Markdown
Owner Author

Research Team — final review: APPROVED (Research)

Verified on head 970c63e:

  • Content: only change since the reviewed version is line 3 → **Type:** Design disclosure (cherry-pick of private b5b04ef4, merged via private #375). File blob 0bbd24e6 identical to private master.
  • IP check: clear (no out-of-scope concepts; thresholds symbolic; standard disclaimer only).
  • Checks: all Actions workflows on this head completed successfully (CI incl. tests 3.10/3.11/3.12 and smoke tests, IP Content Gate, IP Gate, Trade-Secret Wheel Gate, Governance Gate, ADR Visibility Gate, Path-Deny Guard, PR Guardian, graq_predict deployment gate). The publish job is skipped (release-only job; not a test).

Merge and the arXiv submission are the owner's actions.

@harishquantamix
harishquantamix merged commit 4c63724 into master Oct 3, 2026
16 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants