[noahgift]

Third batch fold of branch-only test lines (PRCAP clog): cherry-picks of #5614 (cgp doctor collect_checks) and #5595 (cbtop render_sparklines, own contract cbtop-render-sparklines-v1).

Held out: #5616 adds FALSIFY-OPS-029 to apr-cli-operations-v1, which waits on the OPS id chain. #5726 is superseded by #5724 (identical content).

Derived files regenerated once (pv extract, pv-sat, readme_sync, tree_reader_tests). cgp doctor tests, the cbtop lib suite (624) and contract_data_integrity pass locally.

Refs #5614 #5595
keep-open: each issue keeps its remaining branches; closed by the owner lane
ont-delta: none, tests and their contracts only

🤖 Generated with Claude Code

[noahgift]

seat 1, aprender-pace @d0b1f1f490ddd99e3b262e946589dc38154e2ea1

Read the full diff (+281/-4, 8 files). No production code changed: aprender-cbtop mod.rs only adds '#[cfg(test)] mod sparklines_tests;', aprender-cgp doctor.rs adds tests, plus the new sparklines_tests.rs. Checked against collect_checks on the head: the expected name order (nvidia-smi, CUDA Runtime, ncu, nsys, CUPTI, perf, valgrind, renacer, trueno-explain, GPU, CPU) matches the source, the nvcc->'CUDA Runtime' rename is as asserted, and perf's post-processing only edits version, so the status/path assertion holds for it. Not verified: the tests read the host's PATH (which::which, nvcc, CUPTI paths), so they pass on any host only if check_binary gives Ok exactly when it finds a path; I did not read check_binary. I did not read sparklines_tests.rs line by line. Generated files (README/contracts.nt/witness) not regenerated by me. No tests run, CI not read. 0 aprender-pace trailers.

[noahgift]

seat 1, aprender-triage-worker @d0b1f1f490ddd99e3b262e946589dc38154e2ea1 APPROVE

Read: full diff (8 files, +281/-4). Rust additions are test-only (new sparklines_tests.rs wired by #[cfg(test)] mod, and tests appended in aprender-cgp doctor.rs). Non-addition lines: README count 1984 -> 1986 (2 new contracts, matches), witness census sha + reasoner sha. Boundary arithmetic in the sparkline tests checks out against the contract (start 2 + 4 + 2 = 8; cpu gate y<22, bricks gate y<25). Contracts: cbtop has 1 obligation + 1 falsifier + mutation line; cgp has 2 obligations + 2 falsifiers.
Notes (non-blocking): (1) the cgp test block comment says FALSIFY-CGP-DOCTOR-001..004 but the contract carries 001..002; two tests (cuda_runtime rename, status_matches_path) have no falsifier id. (2) cgp tests are host-dependent by design (which nvcc, CUPTI path) and assert self-consistency, so they hold on any host. (3) README count is absolute: it will conflict with #5723/#5724 (1987/1989) whichever lands second; regenerate, do not hand-merge.
Not run: tests, pv validate, clippy. Trailers: 0 triage-worker.

[github-actions]

§13.11 rung 1 — quorum shadow verdict

S13-SHADOW pr=5727 head=b758d90446a67e9ff53b323c16777d9054e837e8 verdict=REFUSE class=Q1 arm_rc=1

Shadow mode: this records a verdict and merges nothing. A refusal
to arm is not a block (§13 adds zero rows to §7) — the pull request is
exactly as green as it was.

[noahgift]

seat 1, aprender-triage-worker @7be213f782cf2981b327ecc96ecc7b5e7f82814d APPROVE
Re-seat. Delta from my last seat is the contracts YAML formal: lines only, prefixed with "∀ run: " (diffed +/- lines of the old and new PR diff: only those 2 lines differ, tests unchanged). 0 own-author trailers. Not run: build, tests, sigma gate, CI. State is BEHIND; I did not update the branch.

[noahgift]

seat 1, aprender-pace @7be213f782cf2981b327ecc96ecc7b5e7f82814d

Re-seat of my earlier seat @d0b1f1f490dd. Delta in the PR's own diff (excluding contracts.nt, witness, README): two contract 'formal:' strings now start with '∀ run: '; nothing else differs. Sigma formal_prose count not run by me (a3 reports 8/8 locally). Regenerated files not re-inspected; no tests run, CI not read. 0 aprender-pace trailers.