Skip to content

docs(skills): analog-emulation traps from the compile-answer retry (Closes #12637) - #12638

Merged
gHashTag merged 1 commit into
masterfrom
lane/skill-lessons-126
Oct 26, 2026
Merged

gHashTag merged 1 commit into
masterfrom
lane/skill-lessons-126

Conversation

@gHashTag

Copy link
Copy Markdown
Collaborator

What

Three trap lines at the end of "Traps" in .claude/skills/analog-emulation/SKILL.md, from the work of #12539 (PR 12636).

Closes #12637. Refs #10486.

Changes

  • A gate for a change to the tri mutate driver runs with the driver built from the lane; the new summary line (Compile answers ...) shows which driver ran.
  • A rule over a tool's text output goes into a pure function in a sealed spec with recorded answers as tests; the old Rust unit test moves into the spec and pays for the glue under the net-zero budget (34 added, 47 deleted in cli/tri/src/mutate.rs).
  • A path that the live gate run did not exercise is reported with its live count (0 retried) and its unit tests as the only evidence.

Checks run

  • Diff scope: 1 file, 3 added lines, 0 deleted (git diff --stat).
  • Greps on the added lines: 0 non-ASCII bytes, 0 home or tilde paths, 0 hits for person, company and money words.
  • Pre-commit hooks (foreign-budget, own-language, secret-gate) passed on the commit.

Not claimed

No speed claim. The figures quoted (223 killed, 2 survived, 27 hung, 0 unviable, exit 0; 34 added, 47 deleted) are the ones already recorded in the merge basis of PR 12636.

🤖 Generated with Claude Code

…loses #12637)

Three trap lines: the success gate of a driver change runs with the lane-built
driver and the new summary line shows which driver ran; a rule over a tool's
text output lives in a sealed spec and the old unit test pays for the glue;
a path no live run exercised is reported with its live count of 0.

Refs #10486

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-26 06:00:04 UTC

Summary

Status Count
Total Open PRs 22
PRs with Failing Checks 20
PRs with All Checks Green 2
READY 0
FAILING 20
PENDING 0
NO CHECKS YET 1

These columns do not partition: 0 + 20 + 0 + 1 = 21, and there are 22 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=6fb540d220f7 != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@tri-claw

tri-claw Bot commented Oct 26, 2026

Copy link
Copy Markdown

Summary by t27-bees

This PR adds three trap entries to the analog-emulation skill document, documenting lessons learned from PR #12539 about gate execution with lane-built drivers, moving rules to pure functions in sealed specs, and reporting unexercised paths with live counts.

Worth a look

Nothing stood out.


Read: 1 files, +3 -0, the whole diff. Head 135b7a4. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.

@gHashTag

Copy link
Copy Markdown
Collaborator Author

Merge basis (read at 13:02 +07, head 135b7a4eebaebe16aad79be14cf5df4f49af6c22, merge rule of 2026-10-25):

  • Four required checks of t27-master-protection pass: validate, check-linked-issue, parse-ratchet, own-language.
  • Diff (from gh pr view 12638 --json files, the whole PR): 1 file, .claude/skills/analog-emulation/SKILL.md, 3 lines added, 0 deleted. No spec, no generated file, no seal, no Rust.
  • Other checks at the read: 21 pass, 3 skipping, 0 red. Pending: coverage, Corpus ratchet (expected-failure ledger), Verilog widths (absurd-width ledger). A skill text file cannot move a seal, a ledger row or a Verilog width; if one of them ends red in the minute before the merge, it is named in the done comment.
  • Reviews: 0. The dashboard comment is not a review.
  • Local greps on the added lines: 0 non-ASCII bytes, 0 home or tilde paths, 0 hits for person, company and money words.

Merging with --match-head-commit on the head above.

🤖 Addressed by Claude Code

@gHashTag
gHashTag merged commit e104e21 into master Oct 26, 2026
30 of 31 checks passed
@gHashTag

Copy link
Copy Markdown
Collaborator Author

Post-merge read (13:05 +07): coverage on this pull request's head ended red after the merge, in the minute after the basis was read (it was pending at the merge). Run 40277274532 prints "FAIL: 11 seal(s) newly do not hold" and lists ten [stale] seals: cli_git_commands, crons_t27os-night-loop, git_commands, t27b_T27bWasm, t27f_T27fWorkGraph, t27f_a11y_mirror, t27f_hud, t27os_T27osHiveFeedAcceptance, t27os_T27osLabLegs, t27os_labs. All ten are in the set of 11 that master runs 40271326185 and 40273324519 already showed red; none is a seal of a file this pull request touched (it changed one skill text file and no spec). So the red is inherited and independent of the diff. Corpus ratchet was still pending at this read.

🤖 Addressed by Claude Code

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.

docs(skills): analog-emulation traps from the compile-answer retry (#12539)

1 participant