From 50ea2a0f3c32972b16681a5b9848839d6da2f46b Mon Sep 17 00:00:00 2001 From: Jon Langevin Date: Sat, 25 Apr 2026 02:52:56 -0400 Subject: [PATCH 1/8] feat(cross-llm): rewrite Group II skills for cross-host portability MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit alignment-check (Task 4): - Replace "Dispatch a Sonnet agent" with "Dispatch a balanced-tier subagent" - Wrap Agent tool code block in / - model: sonnet → model: balanced pr-monitoring (Task 5): - Replace "background agent" with "background balanced-tier agent" in prose - Wrap entire Agent tool code block in / - model: sonnet → model: balanced brainstorming (Task 8): - Replace inline AskUserQuestion references with host-neutral phrasing - Wrap AskUserQuestion-specific bullet points in - Add fallback: numbered list + 3 questions/turn receiving-code-review (Task 6) and finishing-a-development-branch (Task 7): - Audited; no forbidden tokens found — no changes required. All changes verified: tests/skill-content-grep.sh PASS Co-Authored-By: Claude Sonnet 4.6 --- skills/alignment-check/SKILL.md | 6 ++++-- skills/brainstorming/SKILL.md | 15 ++++++++++++--- skills/pr-monitoring/SKILL.md | 7 +++++-- 3 files changed, 21 insertions(+), 7 deletions(-) diff --git a/skills/alignment-check/SKILL.md b/skills/alignment-check/SKILL.md index ef63821..bdb03c2 100644 --- a/skills/alignment-check/SKILL.md +++ b/skills/alignment-check/SKILL.md @@ -25,10 +25,11 @@ Invoked automatically by `writing-plans` in autonomous mode. Can also be invoked ## Dispatching the Alignment Agent -Dispatch a Sonnet agent to perform the comparison: +Dispatch a `balanced`-tier subagent to perform the comparison using the following prompt: + ``` -Agent tool (general-purpose, model: sonnet): +Agent tool (general-purpose, model: balanced): description: "Check alignment: design vs plan" prompt: | You are verifying that an implementation plan aligns with its design document. @@ -71,6 +72,7 @@ Agent tool (general-purpose, model: sonnet): **Drift Items:** [list specific items to fix] ``` + ## On FAIL diff --git a/skills/brainstorming/SKILL.md b/skills/brainstorming/SKILL.md index 7b83cee..9b30efc 100644 --- a/skills/brainstorming/SKILL.md +++ b/skills/brainstorming/SKILL.md @@ -58,11 +58,20 @@ digraph brainstorming { **Understanding the idea:** - Check out the current project state first (files, docs, recent commits) -- Ask questions using adaptive batching with AskUserQuestion: - - **First form:** Group 2-4 related questions covering purpose, constraints, scope, and tech choices - - **Follow-ups:** Targeted single questions based on interesting or ambiguous answers from previous forms +- Ask questions using adaptive batching — group related questions to reduce round-trips: + - **First batch:** 2–4 questions covering purpose, constraints, scope, and tech choices + - **Follow-ups:** Targeted single questions based on interesting or ambiguous answers + + - Use multiple choice options when possible (AskUserQuestion supports 2-4 options per question) - AskUserQuestion supports up to 4 questions per form — use this to reduce round-trips + + + + - Present options as a numbered list and ask the user to reply with the chosen number + - Group no more than 3 questions per turn to avoid overloading the chat + + - Focus on understanding: purpose, constraints, success criteria **Exploring approaches:** diff --git a/skills/pr-monitoring/SKILL.md b/skills/pr-monitoring/SKILL.md index 2ecebe7..9cbc284 100644 --- a/skills/pr-monitoring/SKILL.md +++ b/skills/pr-monitoring/SKILL.md @@ -17,10 +17,11 @@ Invoked automatically by `finishing-a-development-branch` in autonomous mode aft ## The Process -Spawn a background agent that monitors the PR in a loop: +Spawn a background `balanced`-tier agent that monitors the PR in a loop. + ``` -Agent tool (general-purpose, model: sonnet, run_in_background: true): +Agent tool (general-purpose, model: balanced, run_in_background: true): description: "Monitor PR #N for CI and reviews" prompt: | You are monitoring PR # on and automatically fixing issues. @@ -88,6 +89,8 @@ Agent tool (general-purpose, model: sonnet, run_in_background: true): ### 4. Wait Between Checks Sleep 60 seconds between check cycles. Do not poll more frequently. +``` + ## Safety Limits From 0ce0141af643522c218afdf4fd4c61ecbe8be651 Mon Sep 17 00:00:00 2001 From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com> Date: Sat, 25 Apr 2026 07:01:26 +0000 Subject: [PATCH 2/8] fix(cross-llm): address review feedback on alignment-check and pr-monitoring alignment-check: - Fix misleading "using the following prompt:" sentence; prompt is inside a block and hidden on other hosts - Add "On Claude Code, use the Agent tool:" header inside host block to make the scoping explicit pr-monitoring: - Fix nested fences bug: inner ```bash fences inside Agent prompt were terminating the outer ``` fence; switch outer fence to 4 backticks - Move Monitor Loop steps outside as host-neutral ## Monitor Loop section so non-Claude-Code hosts see the procedure - Keep full Agent tool invocation (with detailed prompt) inside host block for Claude Code background agent Agent-Logs-Url: https://github.com/GoCodeAlone/claude-superpowers/sessions/023600d7-9f65-420f-8d6e-9ccf6b448388 Co-authored-by: intel352 <77607+intel352@users.noreply.github.com> --- skills/alignment-check/SKILL.md | 4 +- skills/pr-monitoring/SKILL.md | 68 ++++++++++++++++++++++++++++++--- 2 files changed, 65 insertions(+), 7 deletions(-) diff --git a/skills/alignment-check/SKILL.md b/skills/alignment-check/SKILL.md index bdb03c2..b562faa 100644 --- a/skills/alignment-check/SKILL.md +++ b/skills/alignment-check/SKILL.md @@ -25,9 +25,11 @@ Invoked automatically by `writing-plans` in autonomous mode. Can also be invoked ## Dispatching the Alignment Agent -Dispatch a `balanced`-tier subagent to perform the comparison using the following prompt: +Dispatch a `balanced`-tier subagent to perform the comparison. +On Claude Code, use the Agent tool: + ``` Agent tool (general-purpose, model: balanced): description: "Check alignment: design vs plan" diff --git a/skills/pr-monitoring/SKILL.md b/skills/pr-monitoring/SKILL.md index 9cbc284..9bc4c7d 100644 --- a/skills/pr-monitoring/SKILL.md +++ b/skills/pr-monitoring/SKILL.md @@ -17,10 +17,12 @@ Invoked automatically by `finishing-a-development-branch` in autonomous mode aft ## The Process -Spawn a background `balanced`-tier agent that monitors the PR in a loop. +Spawn a background `balanced`-tier agent that monitors the PR in a loop following the Monitor Loop below. -``` +Use the Agent tool to run the monitor in the background: + +```` Agent tool (general-purpose, model: balanced, run_in_background: true): description: "Monitor PR #N for CI and reviews" prompt: | @@ -34,9 +36,7 @@ Agent tool (general-purpose, model: balanced, run_in_background: true): Design doc: Plan doc: - ## Monitor Loop - - Repeat until exit conditions met: + Repeat the Monitor Loop until exit conditions are met: ### 1. Check CI Status @@ -89,9 +89,65 @@ Agent tool (general-purpose, model: balanced, run_in_background: true): ### 4. Wait Between Checks Sleep 60 seconds between check cycles. Do not poll more frequently. -``` +```` +## Monitor Loop + +Repeat until exit conditions met: + +### 1. Check CI Status + +```bash +gh pr checks --json name,state,conclusion +``` + +**If any check fails:** +a. Read the failure logs: `gh run view --log-failed` +b. Identify the root cause +c. Fix the issue in the codebase +d. Run the relevant tests locally to verify +e. Commit and push: + ```bash + git add + git commit -m "fix: address CI failure in " + git push + ``` +f. Wait 60s, re-check + +**Safety:** Max 5 fix attempts per unique CI failure. After 5, comment on the PR: +"Unable to automatically resolve CI failure in after 5 attempts. Manual intervention needed." + +### 2. Check Review Comments + +```bash +gh api repos///pulls//comments --jq '.[] | select(.position != null) | {id, body, path, line: .original_line, user: .user.login}' +gh api repos///pulls//reviews --jq '.[] | select(.state == "CHANGES_REQUESTED") | {id, body, user: .user.login}' +``` + +**If new unresolved comments found:** +a. Read the comment carefully +b. Implement the requested change +c. Run tests to verify +d. Commit and push +e. Reply to the comment: "Addressed in " + +**Safety:** Max 3 revision rounds per review comment. After 3, reply: +"I've attempted to address this feedback but may need clarification. Flagging for manual review." + +### 3. Check Exit Conditions + +Exit when ALL of: +- All CI checks passing (green) +- No unresolved review comments +- No pending "changes requested" reviews + +On exit, report final status. + +### 4. Wait Between Checks + +Sleep 60 seconds between check cycles. Do not poll more frequently. + ## Safety Limits | Limit | Value | On Exceed | From e5f59400a6f1696d27a0a21b243e1cfeeaeb71c4 Mon Sep 17 00:00:00 2001 From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com> Date: Sat, 25 Apr 2026 07:04:25 +0000 Subject: [PATCH 3/8] fix(cross-llm): address second review pass on alignment-check and brainstorming MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit alignment-check: - Move forward/reverse trace instructions and report format outside as host-neutral prose; all hosts now see the full alignment procedure - Slim the claude-code Agent tool prompt down to a minimal wrapper that references the skill instructions, avoiding duplication brainstorming: - Fix batch-size inconsistency: host-neutral "2–4 questions" contradicted the cap of 3 per turn; changed to "2–3 questions" for consistency across all hosts Agent-Logs-Url: https://github.com/GoCodeAlone/claude-superpowers/sessions/6c1dc754-4fe3-4943-aab2-97223c410498 Co-authored-by: intel352 <77607+intel352@users.noreply.github.com> --- skills/alignment-check/SKILL.md | 72 ++++++++++++++++----------------- skills/brainstorming/SKILL.md | 2 +- 2 files changed, 36 insertions(+), 38 deletions(-) diff --git a/skills/alignment-check/SKILL.md b/skills/alignment-check/SKILL.md index b562faa..d67c62f 100644 --- a/skills/alignment-check/SKILL.md +++ b/skills/alignment-check/SKILL.md @@ -25,54 +25,52 @@ Invoked automatically by `writing-plans` in autonomous mode. Can also be invoked ## Dispatching the Alignment Agent -Dispatch a `balanced`-tier subagent to perform the comparison. +Dispatch a `balanced`-tier subagent to verify alignment. The subagent reads both documents and produces an Alignment Report: - -On Claude Code, use the Agent tool: - -``` -Agent tool (general-purpose, model: balanced): - description: "Check alignment: design vs plan" - prompt: | - You are verifying that an implementation plan aligns with its design document. +**Input:** +- Design document: `docs/plans/YYYY-MM-DD--design.md` +- Implementation plan: `docs/plans/YYYY-MM-DD-.md` - ## Design Document - [Read: docs/plans/YYYY-MM-DD--design.md] +**Forward trace (design → plan):** +For each requirement in the design: +- Find the plan task(s) that implement it +- If no task covers it: flag as MISSING - ## Implementation Plan - [Read: docs/plans/YYYY-MM-DD-.md] +**Reverse trace (plan → design):** +For each task in the plan: +- Find the design requirement it satisfies +- If no requirement justifies it: flag as SCOPE CREEP - ## Your Job +**Report format:** - **Forward trace (design → plan):** - For each requirement in the design: - - Find the plan task(s) that implement it - - If no task covers it: flag as MISSING +### Alignment Report - **Reverse trace (plan → design):** - For each task in the plan: - - Find the design requirement it satisfies - - If no requirement justifies it: flag as SCOPE CREEP +**Status:** PASS | FAIL - **Report format:** +**Coverage:** +| Design Requirement | Plan Task(s) | Status | +|---|---|---| +| [requirement] | Task N | ✅ Covered | +| [requirement] | — | ❌ MISSING | - ### Alignment Report +**Scope Check:** +| Plan Task | Design Requirement | Status | +|---|---|---| +| Task N | [requirement] | ✅ Justified | +| Task N | — | ⚠️ SCOPE CREEP | - **Status:** PASS | FAIL +**Drift Items:** [list specific items to fix] - **Coverage:** - | Design Requirement | Plan Task(s) | Status | - |---|---|---| - | [requirement] | Task N | ✅ Covered | - | [requirement] | — | ❌ MISSING | - - **Scope Check:** - | Plan Task | Design Requirement | Status | - |---|---|---| - | Task N | [requirement] | ✅ Justified | - | Task N | — | ⚠️ SCOPE CREEP | + +Dispatch using the Agent tool: - **Drift Items:** [list specific items to fix] +``` +Agent tool (general-purpose, model: balanced): + description: "Check alignment: design vs plan" + prompt: | + You are verifying that an implementation plan aligns with its design document. + Read docs/plans/YYYY-MM-DD--design.md and docs/plans/YYYY-MM-DD-.md, + then produce an Alignment Report following the instructions in the alignment-check skill. ``` diff --git a/skills/brainstorming/SKILL.md b/skills/brainstorming/SKILL.md index 9b30efc..0007aad 100644 --- a/skills/brainstorming/SKILL.md +++ b/skills/brainstorming/SKILL.md @@ -59,7 +59,7 @@ digraph brainstorming { **Understanding the idea:** - Check out the current project state first (files, docs, recent commits) - Ask questions using adaptive batching — group related questions to reduce round-trips: - - **First batch:** 2–4 questions covering purpose, constraints, scope, and tech choices + - **First batch:** 2–3 questions covering purpose, constraints, scope, and tech choices - **Follow-ups:** Targeted single questions based on interesting or ambiguous answers From 29d0f38e02cd828bd1571c23a53f180f03f02be4 Mon Sep 17 00:00:00 2001 From: Jon Langevin Date: Sat, 25 Apr 2026 03:17:32 -0400 Subject: [PATCH 4/8] fix(cross-llm): resolve brainstorming number conflict + remove Monitor Loop duplication MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit brainstorming: - Remove specific count from "First batch" line (was "2–3 questions", conflicting with the "2-4 questions per form" principle elsewhere); now reads "First batch: covers purpose, constraints, scope, and tech choices" pr-monitoring: - Remove standalone ## Monitor Loop section (lines 95-149) that duplicated the host-wrapped Agent tool prompt — single source of truth is inside where the loop runs autonomously - Add block with brief gh polling guidance for non-Claude hosts Co-Authored-By: Claude Sonnet 4.6 --- skills/brainstorming/SKILL.md | 2 +- skills/pr-monitoring/SKILL.md | 57 +++-------------------------------- 2 files changed, 6 insertions(+), 53 deletions(-) diff --git a/skills/brainstorming/SKILL.md b/skills/brainstorming/SKILL.md index 0007aad..f09b4fc 100644 --- a/skills/brainstorming/SKILL.md +++ b/skills/brainstorming/SKILL.md @@ -59,7 +59,7 @@ digraph brainstorming { **Understanding the idea:** - Check out the current project state first (files, docs, recent commits) - Ask questions using adaptive batching — group related questions to reduce round-trips: - - **First batch:** 2–3 questions covering purpose, constraints, scope, and tech choices + - **First batch:** covers purpose, constraints, scope, and tech choices - **Follow-ups:** Targeted single questions based on interesting or ambiguous answers diff --git a/skills/pr-monitoring/SKILL.md b/skills/pr-monitoring/SKILL.md index 9bc4c7d..d2bbdba 100644 --- a/skills/pr-monitoring/SKILL.md +++ b/skills/pr-monitoring/SKILL.md @@ -92,61 +92,14 @@ Agent tool (general-purpose, model: balanced, run_in_background: true): ```` -## Monitor Loop + -Repeat until exit conditions met: +Use your host's equivalent mechanism to periodically poll `gh pr checks ` and +`gh api repos///pulls//comments` in a loop, fixing failures and +responding to review comments until all checks pass and no unresolved comments remain. -### 1. Check CI Status - -```bash -gh pr checks --json name,state,conclusion -``` - -**If any check fails:** -a. Read the failure logs: `gh run view --log-failed` -b. Identify the root cause -c. Fix the issue in the codebase -d. Run the relevant tests locally to verify -e. Commit and push: - ```bash - git add - git commit -m "fix: address CI failure in " - git push - ``` -f. Wait 60s, re-check - -**Safety:** Max 5 fix attempts per unique CI failure. After 5, comment on the PR: -"Unable to automatically resolve CI failure in after 5 attempts. Manual intervention needed." - -### 2. Check Review Comments - -```bash -gh api repos///pulls//comments --jq '.[] | select(.position != null) | {id, body, path, line: .original_line, user: .user.login}' -gh api repos///pulls//reviews --jq '.[] | select(.state == "CHANGES_REQUESTED") | {id, body, user: .user.login}' -``` - -**If new unresolved comments found:** -a. Read the comment carefully -b. Implement the requested change -c. Run tests to verify -d. Commit and push -e. Reply to the comment: "Addressed in " - -**Safety:** Max 3 revision rounds per review comment. After 3, reply: -"I've attempted to address this feedback but may need clarification. Flagging for manual review." - -### 3. Check Exit Conditions - -Exit when ALL of: -- All CI checks passing (green) -- No unresolved review comments -- No pending "changes requested" reviews - -On exit, report final status. - -### 4. Wait Between Checks + -Sleep 60 seconds between check cycles. Do not poll more frequently. ## Safety Limits From 844c6b040bd4971ae173bdafbc6a6c987cdfc6f8 Mon Sep 17 00:00:00 2001 From: Jon Langevin Date: Sat, 25 Apr 2026 03:22:39 -0400 Subject: [PATCH 5/8] fix(cross-llm): remove remaining count inconsistencies in brainstorming + pr-monitoring MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit brainstorming: - Remove "2-4 per form" count from checklist and Key Principles; now reads "group related questions to reduce round-trips; use targeted singles for follow-ups" — no number to conflict with cap of 3 per turn pr-monitoring: - Fix dangling "Monitor Loop below" reference (standalone section was removed in previous commit); now "following the Monitor Loop procedure" Co-Authored-By: Claude Sonnet 4.6 --- skills/brainstorming/SKILL.md | 4 ++-- skills/pr-monitoring/SKILL.md | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/skills/brainstorming/SKILL.md b/skills/brainstorming/SKILL.md index f09b4fc..fcfe20a 100644 --- a/skills/brainstorming/SKILL.md +++ b/skills/brainstorming/SKILL.md @@ -24,7 +24,7 @@ Every project goes through this process. A todo list, a single-function utility, You MUST create a task for each of these items and complete them in order: 1. **Explore project context** — check files, docs, recent commits -2. **Ask clarifying questions** — adaptive batching: group 2-4 related questions per form, follow up with targeted singles +2. **Ask clarifying questions** — adaptive batching: group related questions to reduce round-trips; use targeted singles for follow-ups 3. **Propose 2-3 approaches** — with trade-offs and your recommendation 4. **Present design** — in sections scaled to their complexity, get user approval after each section 5. **Write design doc** — save to `docs/plans/YYYY-MM-DD--design.md` and commit @@ -115,7 +115,7 @@ When the user wants design exploration without execution, they pass `--design-on ## Key Principles -- **Adaptive question batching** - Group 2-4 related questions per form, follow up with targeted singles +- **Adaptive question batching** - Group related questions to reduce round-trips; use targeted singles for follow-ups - **Multiple choice preferred** - Easier to answer than open-ended when possible - **YAGNI ruthlessly** - Remove unnecessary features from all designs - **Explore alternatives** - Always propose 2-3 approaches before settling diff --git a/skills/pr-monitoring/SKILL.md b/skills/pr-monitoring/SKILL.md index d2bbdba..18e4ce0 100644 --- a/skills/pr-monitoring/SKILL.md +++ b/skills/pr-monitoring/SKILL.md @@ -17,7 +17,7 @@ Invoked automatically by `finishing-a-development-branch` in autonomous mode aft ## The Process -Spawn a background `balanced`-tier agent that monitors the PR in a loop following the Monitor Loop below. +Spawn a background `balanced`-tier agent that monitors the PR in a loop following the Monitor Loop procedure. Use the Agent tool to run the monitor in the background: From 14fba4f1a54655a5a56e9482f977986a0cc55c52 Mon Sep 17 00:00:00 2001 From: Jon Langevin Date: Sat, 25 Apr 2026 03:28:37 -0400 Subject: [PATCH 6/8] fix(cross-llm): inline alignment procedure into Agent prompt + indent brainstorming host markers Co-Authored-By: Claude Sonnet 4.6 --- skills/alignment-check/SKILL.md | 34 +++++++++++++++++++++++++++++++-- skills/brainstorming/SKILL.md | 8 ++++---- 2 files changed, 36 insertions(+), 6 deletions(-) diff --git a/skills/alignment-check/SKILL.md b/skills/alignment-check/SKILL.md index d67c62f..4affc11 100644 --- a/skills/alignment-check/SKILL.md +++ b/skills/alignment-check/SKILL.md @@ -69,8 +69,38 @@ Agent tool (general-purpose, model: balanced): description: "Check alignment: design vs plan" prompt: | You are verifying that an implementation plan aligns with its design document. - Read docs/plans/YYYY-MM-DD--design.md and docs/plans/YYYY-MM-DD-.md, - then produce an Alignment Report following the instructions in the alignment-check skill. + + Read docs/plans/YYYY-MM-DD--design.md and docs/plans/YYYY-MM-DD-.md. + + Perform a forward trace (design → plan): + - For each requirement, constraint, and acceptance criterion in the design, find the plan task(s) that implement it. + - If no plan task covers a design item, flag it as MISSING. + + Perform a reverse trace (plan → design): + - For each task in the implementation plan, find the design requirement, constraint, or acceptance criterion it satisfies. + - If no design item justifies a plan task, flag it as SCOPE CREEP. + + Return exactly this report format: + + ### Alignment Report + + **Status:** PASS | FAIL + + **Coverage:** + | Design Requirement | Plan Task(s) | Status | + |---|---|---| + | [requirement] | Task N | ✅ Covered | + | [requirement] | — | ❌ MISSING | + + **Scope Check:** + | Plan Task | Design Requirement | Status | + |---|---|---| + | Task N | [requirement] | ✅ Justified | + | Task N | — | ⚠️ SCOPE CREEP | + + **Drift Items:** [list specific items to fix] + + Set **Status:** to PASS only if every design item is covered and every plan task is justified. Otherwise set it to FAIL. ``` diff --git a/skills/brainstorming/SKILL.md b/skills/brainstorming/SKILL.md index fcfe20a..2da13af 100644 --- a/skills/brainstorming/SKILL.md +++ b/skills/brainstorming/SKILL.md @@ -62,15 +62,15 @@ digraph brainstorming { - **First batch:** covers purpose, constraints, scope, and tech choices - **Follow-ups:** Targeted single questions based on interesting or ambiguous answers - + - Use multiple choice options when possible (AskUserQuestion supports 2-4 options per question) - AskUserQuestion supports up to 4 questions per form — use this to reduce round-trips - + - + - Present options as a numbered list and ask the user to reply with the chosen number - Group no more than 3 questions per turn to avoid overloading the chat - + - Focus on understanding: purpose, constraints, success criteria From 64ad84403bf32eb1763e0ba4f16fbe19047c13bf Mon Sep 17 00:00:00 2001 From: Jon Langevin Date: Sat, 25 Apr 2026 03:31:05 -0400 Subject: [PATCH 7/8] fix(cross-llm): rephrase pr-monitoring intro + add CHANGES_REQUESTED to non-Claude host guidance Co-Authored-By: Claude Sonnet 4.6 --- skills/pr-monitoring/SKILL.md | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/skills/pr-monitoring/SKILL.md b/skills/pr-monitoring/SKILL.md index 18e4ce0..d1b66c0 100644 --- a/skills/pr-monitoring/SKILL.md +++ b/skills/pr-monitoring/SKILL.md @@ -17,7 +17,7 @@ Invoked automatically by `finishing-a-development-branch` in autonomous mode aft ## The Process -Spawn a background `balanced`-tier agent that monitors the PR in a loop following the Monitor Loop procedure. +Run a `balanced`-tier agent that monitors the PR in a loop following the Monitor Loop procedure. Use the Agent tool to run the monitor in the background: @@ -94,9 +94,12 @@ Agent tool (general-purpose, model: balanced, run_in_background: true): -Use your host's equivalent mechanism to periodically poll `gh pr checks ` and -`gh api repos///pulls//comments` in a loop, fixing failures and -responding to review comments until all checks pass and no unresolved comments remain. +Use your host's equivalent mechanism to periodically poll the following in a loop: +- `gh pr checks ` — fix any failing CI checks +- `gh api repos///pulls//comments` — respond to inline review comments +- `gh api repos///pulls//reviews` — handle any "CHANGES_REQUESTED" reviews + +Continue until all checks pass, no unresolved inline comments remain, and no "changes requested" reviews are pending. From 8dc74492eaeb857c38390d53ae32d91a7564c30e Mon Sep 17 00:00:00 2001 From: Jon Langevin Date: Sat, 25 Apr 2026 03:36:06 -0400 Subject: [PATCH 8/8] fix(cross-llm): remove dangling Monitor Loop ref + add non-Claude alignment dispatch + fix brainstorming one-at-a-time Co-Authored-By: Claude Sonnet 4.6 --- skills/alignment-check/SKILL.md | 4 ++++ skills/brainstorming/SKILL.md | 2 +- skills/pr-monitoring/SKILL.md | 2 +- 3 files changed, 6 insertions(+), 2 deletions(-) diff --git a/skills/alignment-check/SKILL.md b/skills/alignment-check/SKILL.md index 4affc11..343dc85 100644 --- a/skills/alignment-check/SKILL.md +++ b/skills/alignment-check/SKILL.md @@ -104,6 +104,10 @@ Agent tool (general-purpose, model: balanced): ``` + +Run the alignment check inline: read both documents, perform the forward and reverse traces using the Comparison Procedure above, and produce the Alignment Report. + + ## On FAIL Feed drift items back to `writing-plans` for revision: diff --git a/skills/brainstorming/SKILL.md b/skills/brainstorming/SKILL.md index 2da13af..de249ad 100644 --- a/skills/brainstorming/SKILL.md +++ b/skills/brainstorming/SKILL.md @@ -9,7 +9,7 @@ description: "You MUST use this before any creative work - creating features, bu Help turn ideas into fully formed designs and specs through natural collaborative dialogue. -Start by understanding the current project context, then ask questions one at a time to refine the idea. Once you understand what you're building, present the design and get user approval. +Start by understanding the current project context, then ask questions using adaptive batching to refine the idea. Once you understand what you're building, present the design and get user approval. Do NOT invoke any implementation skill, write any code, scaffold any project, or take any implementation action until you have presented a design and the user has approved it. This applies to EVERY project regardless of perceived simplicity. diff --git a/skills/pr-monitoring/SKILL.md b/skills/pr-monitoring/SKILL.md index d1b66c0..eef5374 100644 --- a/skills/pr-monitoring/SKILL.md +++ b/skills/pr-monitoring/SKILL.md @@ -17,7 +17,7 @@ Invoked automatically by `finishing-a-development-branch` in autonomous mode aft ## The Process -Run a `balanced`-tier agent that monitors the PR in a loop following the Monitor Loop procedure. +Run a `balanced`-tier agent that monitors the PR in a loop until all CI checks pass and no unresolved reviews remain. Use the Agent tool to run the monitor in the background: