…der syntax + team-lead graph label

- Switch {N}/{team-name} curly-brace placeholders to <N>/<team-name> angle-bracket
  style in implementer-prompt.md and spec-reviewer-prompt.md, consistent with
  all other skill templates
- Update DOT graph node label from "Team Lead" to "team-lead" to match the
  SendMessage recipient identifier used throughout the workflow text
- Add fixture invariant #6 + verify command for curly-brace placeholder detection

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* skill(subagent-driven-development): DRY team conventions to agents/team-conventions.md

Extract repeated boilerplate from implementer/spec-reviewer/code-reviewer
prompts into a single shared file. Each prompt template now references
agents/team-conventions.md instead of inlining TDD discipline, self-review
checklist, review protocol, and sanitization rules.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* skill(subagent-driven-development): fixture — DRY conventions before/after

Documents the before/after of extracting team conventions from inline
prompt boilerplate to agents/team-conventions.md.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address 5 Copilot round-1 findings

A: conventions.md code-reviewer — bug-class checklist + verdict vocabulary
   now reference skills/requesting-code-review/SKILL.md explicitly.
B: conventions.md implementer — version-skew audit now references
   skills/finishing-a-development-branch/SKILL.md Step 1c.
C: implementer-prompt.md — same version-skew audit Step 1c reference.
D: code-quality-reviewer-prompt.md — split adversarial framing from
   bug-class checklist + verdict vocabulary; latter now point to
   skills/requesting-code-review/SKILL.md.
E: conventions.md code-reviewer — adversarial framing exception clause
   preserved: if fewer than 3 issues found, document every bug-class
   check run rather than manufacturing issues.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address 2 Copilot round-2 findings

A: implementer-prompt.md — restore explicit "DM spec-reviewer when ready"
   step so the spec-compliance gate isn't stalled after implementation.
B: SKILL.md — fix stale "checklist below" reference in implementer workflow
   step 4; now points to agents/team-conventions.md.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address 2 Copilot round-3 findings

A: implementer-prompt.md — DM team-lead with branch/commit + "ready for
   merge" (not PR link; PR is created by team-lead via finishing-a-branch
   after all tasks complete, not by the implementer).
B: SKILL.md code-reviewer section — split bug-class checklist + verdict
   vocabulary to their canonical source (skills/requesting-code-review/
   SKILL.md); team-conventions only hosts adversarial framing + output format.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address round-4 finding + tighten fixture

SKILL.md implementer Team Conventions block: replace enumeration that
implied "self-review checklist" is inline with explicit "all rules are
defined in agents/team-conventions.md, not repeated here"; also sync
version-skew audit pointer to finishing-a-development-branch Step 1c.

Fixture: add 5 invariants the "after" state must hold, including:
- no "checklist below" / "see below" / "inlined here" language
- every version-skew audit reference points to Step 1c
- every bug-class / verdict reference points to requesting-code-review
- adversarial-framing exception clause preserved verbatim
Verify command excludes test-fixtures dir to avoid self-matching.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address 2 Copilot round-5 findings

A: fixture verify command — switch to grep -rEn with ERE; replace \b
   (non-portable) with ([[:space:][:punct:]]|$) for reliable Self-Review
   boundary match across grep implementations.
B: team-conventions.md sanitization rule — tighten "no version references"
   to "no specific internal project/company/product-version/incident
   references"; explicitly allow dependency/runtime/tooling version numbers
   needed for accurate technical guidance (version-skew audit, pin refs).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address round-6 finding + fixture ref check

code-quality-reviewer-prompt.md: fix broken path "requesting-code-review/
code-reviewer.md" → "skills/requesting-code-review/code-reviewer.md"
(missing skills/ prefix; file exists at the correct path).

Fixture: add 2 new invariant checks:
- Dead skill file refs: grep all skills/*.md references, verify each
  resolves to an existing file (catches missing-prefix bugs like this one).
- Bare requesting-code-review refs: grep for refs without skills/ prefix
  (both checks exclude test-fixtures dir to avoid self-matching).
All three invariant checks PASS on current tree.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): normalize team-lead spelling across all role prompts

Round-7 fix: use hyphenated team-lead consistently with agents/team-conventions.md
and implementer-prompt.md usage pattern, avoiding ambiguity with the Agent Teams
role identifier.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address round-9 findings — placeholder syntax + team-lead graph label

- Switch {N}/{team-name} curly-brace placeholders to <N>/<team-name> angle-bracket
  style in implementer-prompt.md and spec-reviewer-prompt.md, consistent with
  all other skill templates
- Update DOT graph node label from "Team Lead" to "team-lead" to match the
  SendMessage recipient identifier used throughout the workflow text
- Add fixture invariant #6 + verify command for curly-brace placeholder detection

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): fix fixture grep to catch start-of-line bare refs

Round-10 finding: [^/]requesting-code-review/ requires a preceding character,
so a bare ref at the start of a line would be missed. Switch to
(^|[^/])requesting-code-review/ to cover both cases.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): remove parenthetical rule enumeration — point to team-conventions only

Round-11 findings: both implementer-prompt.md and the embedded implementer
prompt in SKILL.md enumerate convention details (TDD, regression-invariant,
etc.) that are already centralized in agents/team-conventions.md. This
re-inlining contradicts the DRY goal of the refactor. Trim to a simple
pointer; the single source of truth is the conventions file.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(subagent-driven-development): address round-12 findings — PLAN_REFERENCE field + grammar

- Add PLAN_REFERENCE field to code-quality-reviewer dispatch template to match
  the {PLAN_REFERENCE} placeholder in skills/requesting-code-review/code-reviewer.md
  (distinct from PLAN_OR_REQUIREMENTS used in the instructions section)
- Fix grammar in agents/team-conventions.md: "Shared rules every team agent"
  → "Shared rules that every team agent"

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

… plan+design

Finding #1 (plan:395): verification command pipes to grep, which exits 1 when no
matches — looks like failure. Fix: add || true, update expected output description.

Finding #2 (design:142): described markers as "HTML-comment-shaped" but the
syntax is angle-bracket tags (not HTML comments). Fix: "angle-bracket tag syntax
that most markdown renderers treat as unknown HTML tags".

Finding #3 (design:178): prose said OpenCode/Cursor "get blank entries" but
table uses sentinel value host-pass-through. Fix: prose now matches sentinel.

Finding #4 (plan:1232): "Two distinct clean Copilot review timestamps" hard gate
has no anchor policy doc. Fix: downgrade to "recommended practice" phrasing.

Finding #5 (plan:65): skill-content-grep.expected.txt listed as a created file
and in ALLOWED_FILES but the script never reads it — dead artifact. Fix: drop
the file from the task entirely.

Finding #6 (plan:141): guard script ran grep twice per token, reported line
numbers from the stripped (AWK-processed) string rather than the original file.
Fix: single-pass AWK that emits "LINENO:content" preserving original line
numbers; single grep pass with || true.

Finding #7 (plan:132): skip-regex <host: *claude-code *> only matched the exact
single-host tag form, not comma-separated lists like <host: codex, claude-code>.
Fix: AWK rule checks if /claude-code/ appears anywhere in the <host:> line,
handling any comma-separated host list.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

… plan+design

Finding #1 (plan:395): verification command pipes to grep, which exits 1 when no
matches — looks like failure. Fix: add || true, update expected output description.

Finding #2 (design:142): described markers as "HTML-comment-shaped" but the
syntax is angle-bracket tags (not HTML comments). Fix: "angle-bracket tag syntax
that most markdown renderers treat as unknown HTML tags".

Finding #3 (design:178): prose said OpenCode/Cursor "get blank entries" but
table uses sentinel value host-pass-through. Fix: prose now matches sentinel.

Finding #4 (plan:1232): "Two distinct clean Copilot review timestamps" hard gate
has no anchor policy doc. Fix: downgrade to "recommended practice" phrasing.

Finding #5 (plan:65): skill-content-grep.expected.txt listed as a created file
and in ALLOWED_FILES but the script never reads it — dead artifact. Fix: drop
the file from the task entirely.

Finding #6 (plan:141): guard script ran grep twice per token, reported line
numbers from the stripped (AWK-processed) string rather than the original file.
Fix: single-pass AWK that emits "LINENO:content" preserving original line
numbers; single grep pass with || true.

Finding #7 (plan:132): skip-regex <host: *claude-code *> only matched the exact
single-host tag form, not comma-separated lists like <host: codex, claude-code>.
Fix: AWK rule checks if /claude-code/ appears anywhere in the <host:> line,
handling any comma-separated host list.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

…ign docs (#13)

* fix(pr12-round2): address 7 Copilot findings on cross-LLM portability plan+design

Finding #1 (plan:395): verification command pipes to grep, which exits 1 when no
matches — looks like failure. Fix: add || true, update expected output description.

Finding #2 (design:142): described markers as "HTML-comment-shaped" but the
syntax is angle-bracket tags (not HTML comments). Fix: "angle-bracket tag syntax
that most markdown renderers treat as unknown HTML tags".

Finding #3 (design:178): prose said OpenCode/Cursor "get blank entries" but
table uses sentinel value host-pass-through. Fix: prose now matches sentinel.

Finding #4 (plan:1232): "Two distinct clean Copilot review timestamps" hard gate
has no anchor policy doc. Fix: downgrade to "recommended practice" phrasing.

Finding #5 (plan:65): skill-content-grep.expected.txt listed as a created file
and in ALLOWED_FILES but the script never reads it — dead artifact. Fix: drop
the file from the task entirely.

Finding #6 (plan:141): guard script ran grep twice per token, reported line
numbers from the stripped (AWK-processed) string rather than the original file.
Fix: single-pass AWK that emits "LINENO:content" preserving original line
numbers; single grep pass with || true.

Finding #7 (plan:132): skip-regex <host: *claude-code *> only matched the exact
single-host tag form, not comma-separated lists like <host: codex, claude-code>.
Fix: AWK rule checks if /claude-code/ appears anywhere in the <host:> line,
handling any comma-separated host list.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(cross-llm-plan): address 14 Copilot round-1 findings on plan+design

Plan doc (cross-llm-portability.md):
- Add lowercase sonnet/opus/haiku to forbidden-tokens list and explain why
- Add rationale for not yet including AskUserQuestion/Agent (PRs B-D)
- Remove tests/skill-content-grep.sh from ALLOWED_FILES (tests/ not scanned)
- Fix AWK skip to match only exclusive <host: claude-code> blocks, not
  multi-host tags; update inline comment accordingly
- Remove dead fail=1 inside while-pipe subshell; rely solely on $tmp check
- Add || true to all 7 skill-scoped grep verification commands (lines
  443, 517, 626, 694, 764, 844, 888) — grep exits 1 on no-match (PASS case)

Design doc (cross-llm-portability-design.md):
- Fix broken path: subagent-driven-development.md →
  skills/subagent-driven-development/SKILL.md
- Remove false gpt-5.x claim from grep guard description; clarify that
  only Claude-Code brand names are guarded, Codex names use <host:> markers
- Replace "Open questions deferred to implementation" with resolved decisions:
  AWK guard approach, angle-bracket marker syntax, OpenCode tool mapping,
  Cursor support depth scope

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(cross-llm-plan): address 7 Copilot round-2 findings on plan + design docs

- set -euo pipefail (was set -u only)
- Fix stray || true in step-title prose and GHA step name
- Verification commands capture guard output before grepping to avoid masking execution errors
- Fix prose claiming gpt-5.x tokens are guarded (guard only covers Claude brand names)
- Design doc: clarify angle-bracket marker advantage is readability in rendered Markdown

* fix(cross-llm-plan): add end-anchor to AWK skip regex so multi-host blocks are never skipped

Agent-Logs-Url: https://github.com/GoCodeAlone/claude-superpowers/sessions/24d5a30f-f7b3-43e7-83dd-60c470d0f4a4

Co-authored-by: intel352 <77607+intel352@users.noreply.github.com>

* Update docs/plans/2026-04-25-cross-llm-portability.md

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: intel352 <77607+intel352@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>