plan(D-RPF): synergies with the 2026-10-08 probes; separable eligibility

[AdaWorldAPI]

This PR changes only the plan (.claude/plans/2026-10-08-resident-projection-fold-mask-v1.md); there is no code change.

Correction (checked against the code)

D-RPF-1 said recipe eligibility is an Or of up to 24 × 8 = 192 patterns. That is wrong. measure_state computes state[raw5] & pearl[pearl3] (affordance_law.rs:267-270), so each recipe's mask splits into two parts: Or(S_r) ∧ Or(P_r). That is at most 24 + 8 patterns, and fewer once each set is minimised.

The difference matters for cost. D-RPF-9 measured about one full row pass, roughly 1 ms per 64K rows, for each strided predicate. So 192 patterns would cost about 0.2 s and 32 about 32 ms. The reserved codes 24..31 stay excluded rather than treated as "don't care": decode refuses them, and a refusal is not the same as "ineligible".

Synergy table (working model, nothing implemented)

The table maps each finding from D-RPF-9, D-MHB-1/2, D-ART-1 and D-PHT-1 onto the D-RPF phase it affects:

• R2 pattern merge applies to Epi5 ∧ Pearl3: both sit in the same edge word and their bits never overlap.
• The admission plane saves no time until a strided _under kernel (G1) exists. Its classid read shares a 64-byte line with window 0.
• Gate and tile-skip choice should use an exact popcount per query. RollingFloor stays parked.
• D-RPF-4's missing LUT-weighted sum has an exact formulation using D-MHB-1's bit-sliced fold. Quantise the weight table, then count rows over the C-byte pattern set for each bit. It is slow but exact, which makes it a reference oracle rather than the fast path. The c = 255 revision defect must become a stated LUT entry.
• D-RPF-3's dedup key should be looked up in a static array, not a hash map, and it keeps D-ART-1's rule about transforms and group keys.

Before D-RPF-1 ships, it still needs to measure the minimised cover size per recipe on the real law tables.

🤖 Generated with Claude Code

https://claude.ai/code/session_01R2EuNHrariL6B3G3agyASK

---

Generated by Claude Code

[coderabbitai]

Warning

Review limit reached

Your organization has reached its usage spending cap. Adjust your spending cap in the billing tab.

Next included review available in 21 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available. Your 65 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Essentials
• Run ID: 70aadd73-88ed-4a1a-b20a-70ef7d1dc3f5

📥 Commits

Reviewing files that changed from the base of the PR and between e9d1142 and 88c6bc0.

📒 Files selected for processing (1)

• .claude/plans/2026-10-08-resident-projection-fold-mask-v1.md

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

---

_{Comment @coderabbitai help to get the list of available commands.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-08T20:28:44.514788Z	88c6bc0	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.