Skip to content

skill(issue-shape): how to file an issue the shape gate keeps open (Closes #10060) - #10063

Merged
gHashTag merged 1 commit into
masterfrom
skill/issue-shape-10060
Oct 17, 2026
Merged

gHashTag merged 1 commit into
masterfrom
skill/issue-shape-10060

Conversation

@gHashTag

Copy link
Copy Markdown
Owner

Closes #10060

What

The issue-shape gate (specs/policy/issue_shape.t27, #7498, workflow needs-boundary.yml) closed #10054 seconds after filing because the body had no Boundary, User Scenarios or Requirements sections. Nothing in AGENTS.md or the skills mentioned the gate. This PR adds pointers to the spec, which stays the single source of truth:

  • .claude/skills/issue-shape/SKILL.md: a reading guide to the spec, which wins wherever the two disagree. It covers:
    • the four sections and their exact English headings, plus the path :: fn NAME region form;
    • the Boundary trap: every path token inside the section is reserved, so prose that mentions paths goes above it;
    • checking a body with the generated check_issue() on the Railway lab before filing, using the harness copied from the workflow, with a negative control;
    • recovering from shape-refused with gh issue edit --body-file; a passing edit makes the workflow reopen the issue and remove the label.
  • specs/skills/t27-issue-shape.t27: the skill card, following the specs/skills/README.md schema.
  • AGENTS.md section 3: one pointer line under the existing "Issue gate" item. It does not restate the checks.

No gate logic changes.

Evidence (Railway lab, t27c release build)

t27c test-report specs/skills/t27-issue-shape.t27
  tests 1  pass 1  FAIL 0  rate 100.0%
  10 runtime asserts executed (basic_constants), vacuous passes 0 of 1
EXIT=0

Negative control: a copy that asserts ID == "t27/wrong" reports FAIL basic_constants.

The new text is ASCII. The local pre-commit own-language and foreign-budget hooks pass: this PR adds only prose and a .t27 file.

Noticed, not changed

The header comment in specs/policy/issue_shape.t27 names the workflow .github/workflows/issue-shape.yml, but the file in the tree is needs-boundary.yml. The skill says so; the spec comment is left for a separate change.

🤖 Generated with Claude Code

…loses #10060)

The issue-shape gate (specs/policy/issue_shape.t27, #7498) closed #10054
seconds after filing because the body lacked Boundary, User Scenarios and
Requirements. Nothing in AGENTS.md or the skills told an agent about it.

- .claude/skills/issue-shape/SKILL.md: reading guide to the spec (the spec
  wins): sections, Boundary path trap, checking a body with the generated
  check_issue() on the lab, recovery from shape-refused via gh issue edit.
- specs/skills/t27-issue-shape.t27: skill card per specs/skills/README.md.
- AGENTS.md section 3: a pointer next to the Issue gate item, no restated
  checks.

Lab: t27c test-report specs/skills/t27-issue-shape.t27 -> 1 pass, 10 runtime
asserts, exit 0; a wrong-ID copy fails (negative control).

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

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-17 17:52:23 UTC

Summary

Status Count
Total Open PRs 50
PRs with Failing Checks 49
PRs with All Checks Green 1
READY 0
FAILING 49
PENDING 0
NO CHECKS YET 0

These columns do not partition: 0 + 49 + 0 + 0 = 49, and there are 50 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=e769ef021713 != 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).

@gHashTag
gHashTag enabled auto-merge (squash) October 17, 2026 17:56
@gHashTag
gHashTag merged commit 063da04 into master Oct 17, 2026
25 of 29 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

1 participant