Skip to content

fix(commands): add explicit EnterPlanMode/ExitPlanMode to execute, quick, settings #182

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
maystudios merged 1 commit into from
Mar 25, 2026
Merged

fix(commands): add explicit EnterPlanMode/ExitPlanMode to execute, quick, settings #182

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 9 additions & 6 deletions templates/commands/maxsim/execute.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,10 +22,13 @@ Re-entry: If phase is already executed and verified, show status and offer optio
<process>
Follow @.claude/maxsim/workflows/execute.md end-to-end.

1. Detect phase state from GitHub Project Board (already done / partially executed / ready)
2. Group task Issues by wave — re-evaluate wave composition before each spawn against current GitHub state, execute parallel Agents within each wave, sequential across waves
3. Each Agent updates its GitHub Issue label on start and completion
4. After all tasks complete, spawn a verifier Agent to check acceptance criteria
5. On verification failure, auto-retry with gap-closure context (max 3 retries, 4 total attempts)
6. On final failure, report what failed and surface options to user
1. **Plan Mode:** Call `EnterPlanMode` before any execution
2. Detect phase state from GitHub Project Board (already done / partially executed / ready)
3. Group task Issues by wave — re-evaluate wave composition before each spawn against current GitHub state. Present the execution plan to the user for review.
4. Exit Plan Mode via `ExitPlanMode` — user reviews and approves the execution plan
Comment on lines +25 to +28
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

templates/workflows/execute.md only enters Plan Mode in the “Needs execution” path (after detecting state) and can skip it entirely for the “Already executed and verified” re-entry flow. This template’s step 1 (“EnterPlanMode before any execution”) reads as unconditional; consider making it conditional on actually executing (and keep the re-entry flow outside Plan Mode) so it mirrors the workflow’s control flow.

Suggested change
1. **Plan Mode:** Call `EnterPlanMode` before any execution
2. Detect phase state from GitHub Project Board (already done / partially executed / ready)
3. Group task Issues by wave — re-evaluate wave composition before each spawn against current GitHub state. Present the execution plan to the user for review.
4. Exit Plan Mode via `ExitPlanMode` — user reviews and approves the execution plan
1. **Plan Mode (when execution is required):** If the detected phase state indicates that execution is needed, call `EnterPlanMode` before performing any execution steps.
2. Detect phase state from GitHub Project Board (already done / partially executed / ready)
3. Group task Issues by wave — re-evaluate wave composition before each spawn against current GitHub state. Present the execution plan to the user for review.
4. If you entered Plan Mode, exit Plan Mode via `ExitPlanMode` — user reviews and approves the execution plan. The "already executed and verified" re-entry flow stays outside Plan Mode.

Copilot uses AI. Check for mistakes.
Comment on lines +27 to +28
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The description of ExitPlanMode is misleading here: per templates/workflows/execute.md, you present the execution plan and obtain user confirmation in Plan Mode, then call ExitPlanMode after confirmation before spawning executors. Please reword/add an explicit approval step so the Plan Mode lifecycle matches the workflow and doesn’t imply approval happens after exiting.

Suggested change
3. Group task Issues by wave — re-evaluate wave composition before each spawn against current GitHub state. Present the execution plan to the user for review.
4. Exit Plan Mode via `ExitPlanMode` — user reviews and approves the execution plan
3. Group task Issues by wave — re-evaluate wave composition before each spawn against current GitHub state. While still in Plan Mode, present the execution plan to the user for review and obtain explicit approval.
4. After the user approves the plan, exit Plan Mode via `ExitPlanMode` before spawning any Agents

Copilot uses AI. Check for mistakes.
5. Execute parallel Agents within each wave, sequential across waves
6. Each Agent updates its GitHub Issue label on start and completion
7. After all tasks complete, spawn a verifier Agent to check acceptance criteria
8. On verification failure, auto-retry with gap-closure context (max 3 retries, 4 total attempts)
9. On final failure, report what failed and surface options to user
</process>
20 changes: 11 additions & 9 deletions templates/commands/maxsim/quick.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,13 +20,15 @@ Quick tasks are tracked as GitHub Issues (label: `type:quick`) — separate from
<process>
Follow @.claude/maxsim/workflows/quick.md end-to-end.

1. Get task description from $ARGUMENTS or via AskUserQuestion
2. Clarify scope if ambiguous (one focused question)
3. Create a GitHub Issue labeled `type:quick` for the task
4. Spawn a planner Agent (quick mode) to produce a concise implementation plan
5. Spawn executor Agent(s) to implement the plan
6. Spawn a verifier Agent to check the result (tests pass, build succeeds, lint clean)
7. If verification fails, spawn a fix agent (max 2 retries)
8. Commit with atomic message referencing the GitHub Issue
9. Close the GitHub Issue with completion summary
1. **Plan Mode:** Call `EnterPlanMode` before any planning or execution
2. Get task description from $ARGUMENTS or via AskUserQuestion
3. Clarify scope if ambiguous (one focused question)
4. Create a GitHub Issue labeled `type:quick` for the task
5. Spawn a planner Agent (quick mode) to produce a concise implementation plan
6. Present the plan to user — Exit Plan Mode via `ExitPlanMode`
7. Spawn executor Agent(s) to implement the plan
8. Spawn a verifier Agent to check the result (tests pass, build succeeds, lint clean)
9. If verification fails, spawn a fix agent (max 2 retries)
10. Commit with atomic message referencing the GitHub Issue
11. Close the GitHub Issue with completion summary
Comment on lines +23 to +33
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In this command template, the high-level process list no longer matches the referenced workflow (templates/workflows/quick.md). The workflow does not spawn a planner agent or a fix agent/retry loop; it enters Plan Mode only when presenting the plan (after issue creation/model resolution) and exits after explicit user approval. Please align these steps to the workflow so the command doesn’t instruct a different orchestration than the implementation docs it points at.

Suggested change
1. **Plan Mode:** Call `EnterPlanMode` before any planning or execution
2. Get task description from $ARGUMENTS or via AskUserQuestion
3. Clarify scope if ambiguous (one focused question)
4. Create a GitHub Issue labeled `type:quick` for the task
5. Spawn a planner Agent (quick mode) to produce a concise implementation plan
6. Present the plan to user — Exit Plan Mode via `ExitPlanMode`
7. Spawn executor Agent(s) to implement the plan
8. Spawn a verifier Agent to check the result (tests pass, build succeeds, lint clean)
9. If verification fails, spawn a fix agent (max 2 retries)
10. Commit with atomic message referencing the GitHub Issue
11. Close the GitHub Issue with completion summary
1. Get task description from $ARGUMENTS or via AskUserQuestion
2. Clarify scope if ambiguous (one focused question)
3. Create a GitHub Issue labeled `type:quick` for the task
4. Resolve any required model or tool choices as described in the quick workflow
5. Produce a concise implementation plan for the task
6. Enter Plan Mode via `EnterPlanMode` and present the plan to the user for approval
7. After explicit user approval, Exit Plan Mode via `ExitPlanMode`
8. Implement the approved plan (edits, tests, etc.)
9. Run verification checks (e.g., tests, build, lint) as described in the quick workflow
10. Commit with an atomic message referencing the GitHub Issue
11. Close the GitHub Issue with a completion summary

Copilot uses AI. Check for mistakes.
Comment on lines +28 to +33
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The wording around ExitPlanMode is backwards here: the user review/approval should happen while still in Plan Mode, and ExitPlanMode should be called only after the user has approved. Consider adding an explicit “wait for approval” step and describing ExitPlanMode as the transition after approval (matching templates/workflows/quick.md).

Suggested change
6. Present the plan to user — Exit Plan Mode via `ExitPlanMode`
7. Spawn executor Agent(s) to implement the plan
8. Spawn a verifier Agent to check the result (tests pass, build succeeds, lint clean)
9. If verification fails, spawn a fix agent (max 2 retries)
10. Commit with atomic message referencing the GitHub Issue
11. Close the GitHub Issue with completion summary
6. Present the plan to the user and wait for explicit approval while remaining in Plan Mode
7. After the user approves the plan, transition out of Plan Mode by calling `ExitPlanMode`
8. Spawn executor Agent(s) to implement the approved plan
9. Spawn a verifier Agent to check the result (tests pass, build succeeds, lint clean)
10. If verification fails, spawn a fix agent (max 2 retries)
11. Commit with atomic message referencing the GitHub Issue
12. Close the GitHub Issue with completion summary

Copilot uses AI. Check for mistakes.
</process>
13 changes: 8 additions & 5 deletions templates/commands/maxsim/settings.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,16 +16,19 @@ Configuration is stored in `.claude/maxsim/config.json` (project-level). Present
<process>
Follow @.claude/maxsim/workflows/settings.md end-to-end.

1. Read current config from CLAUDE.md and `.claude/maxsim/config.json`
2. Display current settings with descriptions
3. Use AskUserQuestion to interactively configure:
1. **Plan Mode:** Call `EnterPlanMode` before presenting settings
2. Read current config from CLAUDE.md and `.claude/maxsim/config.json`
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This step says to read config from CLAUDE.md, but the referenced workflow (templates/workflows/settings.md) loads config via maxsim-tools (config-ensure-section/config-get) and does not use CLAUDE.md as a config source. To avoid conflicting instructions, update the command template to match the workflow’s actual config-loading mechanism (and mention CLAUDE.md only if it’s truly required).

Suggested change
2. Read current config from CLAUDE.md and `.claude/maxsim/config.json`
2. Use the workflow’s `maxsim-tools` config helpers (e.g. `config-ensure-section`, `config-get`) to load the current config from `.claude/maxsim/config.json`

Copilot uses AI. Check for mistakes.
3. Display current settings with descriptions
4. Use AskUserQuestion to interactively configure:
- Model profile (with per-tier model assignment details)
- Research stage enabled/disabled
- Plan-checker enabled/disabled
- Verifier enabled/disabled
- Auto-advance between stages
- Branching strategy (per-phase / per-task / none)
- Per-agent model overrides (override individual agent models beyond profile defaults)
4. Merge answers and write updated config
5. Display confirmation of saved settings
5. Present proposed configuration changes for review
6. Exit Plan Mode via `ExitPlanMode` — user reviews and approves changes
7. Merge answers and write updated config
Comment on lines +30 to +32
Copy link

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Like the workflow, the user should review/confirm proposed settings while still in Plan Mode; ExitPlanMode should happen only after the user confirms and immediately before writing config.json. The current phrasing (“Exit Plan Mode — user reviews and approves”) suggests approval happens after exiting Plan Mode, which is the opposite of templates/workflows/settings.md and can confuse the operator.

Suggested change
5. Present proposed configuration changes for review
6. Exit Plan Mode via `ExitPlanMode` — user reviews and approves changes
7. Merge answers and write updated config
5. Present proposed configuration changes for review and confirmation (still in Plan Mode)
6. User reviews and approves or adjusts the proposed changes while remaining in Plan Mode
7. After user approval, call `ExitPlanMode` and then merge answers and write the updated config

Copilot uses AI. Check for mistakes.
8. Display confirmation of saved settings
</process>
Loading