/swe-wm-update

Single source of truth for WM updates. Replaces manual REF_WM reads.

This skill ensures comprehensive WM updates with nothing missed. Every WF_* transition MUST use this skill instead of manual updates.

Step 1: Read Current WM

mcp__plugin_swe_serena__read_memory("WM_{session_id}")

Note these daemon-managed fields exactly as they appear (DO NOT MODIFY):

• **Current State**: — daemon updates on WF_* reads
• **Previous State**: — daemon updates automatically
• ### Transitions — daemon appends automatically
• **Edit Count Since Checkpoint**: — daemon increments on file edits
• **Last Updated**: — daemon timestamps automatically

Step 2: Gather Data Using Step-Specific Checklist

Find the checklist matching your --from argument below. Complete ALL items.

WF_CLASSIFY

• Features identified from INDEX_FEATURES
• FEATURE_[KEY] loaded for each feature
• Task type classified (simple / medium / large / operational)
• ### Affected Features populated with Primary / Secondary
• ### Progress updated with classification steps completed

WF_LOAD_FEATURE

• INDEX_FEATURES read
• FEATURE_[KEY] loaded for each feature
• Supporting memories loaded (DOM_, SYS_, REF_*)
• Requirements validated against domain memories (or "none detected")
• **Files:** lists key file paths from feature memories
• ### Progress updated with feature loading outcomes

WF_RESEARCH

• Research findings summarized in ### Notes
• Symbols / patterns discovered noted
• **Files:** lists files examined
• ### Progress updated with research outcomes

WF_CONTINUE

• Previous state verified from WM
• Resume point identified
• Progress carried forward from previous session
• **Files:** lists files from previous work

WF_ARCH_REVIEW

• Design documented with explicit file paths in ### Notes
• **Files:** lists files to modify/create
• Layer compliance verified (pass / fail per criterion)
• Swarm assessment completed (needed / not needed)
• User approval status noted
• ### Progress updated with review results

WF_SWARM_ORCHESTRATE

• Swarm configuration documented in ### Notes
• Agent assignments noted
• Topology selected
• ### Progress updated with orchestration steps

WF_EXECUTE

• Current subtask status
• **Files:** lists files modified / created with descriptions
• Tests written / run status
• Blockers noted (if any)
• ### Progress updated with implementation steps

WF_CHECKPOINT

• All work since last checkpoint summarized
• **Files:** lists files modified with change descriptions
• Current phase / subtask status
• ### Progress updated comprehensively

WF_VERIFY

• CLAUDE_OBLIGATIONS compliance verified
• Architecture compliance verified
• Test coverage verified
• All progress items marked complete or noted
• **Files:** is complete (every file touched in session)
• Status set to [COMPLETED] or [VERIFY_COMPLETE]

WF_DONE

• Status set to [COMPLETED]
• Summary of all work done in ### Context
• All memories updated during session listed in ### Notes
• Follow-up items documented (if any)
• ### Progress — all items checked off

WF_CLARIFY

• Clarification question and user response noted
• Task description updated if scope changed
• ### Progress updated with clarification outcome

Step 3: Write Complete WM

Use ONE write_memory call. Preserve all daemon-managed fields exactly.

mcp__plugin_swe_serena__write_memory("WM_{session_id}", "<complete content>")

Template — Fill ALL Sections

# Working Memory

## Chat: {descriptor}

Session: {session_id}

## Workflow Context

- **Calling Step**: {calling_step}
- **Feature Key(s)**: {feature_keys}
- **Session ID**: {session_id}
- **Return Step**: {return_step}
- **Invocation Mode**: {invocation_mode}
- **Current State**: {PRESERVE FROM STEP 1}
- **Previous State**: {PRESERVE FROM STEP 1}
- **Task Iteration**: {iteration}
- **Edit Count Since Checkpoint:** {PRESERVE FROM STEP 1}
- **Last Updated**: {PRESERVE FROM STEP 1}

### Transitions

{PRESERVE FROM STEP 1}

## Current Task

**[{STATUS}]**: {task_name}

### Context

{1-2 sentence description of current work}

### Feature(s)

{feature key or comma-separated list}

### Affected Features (if multi-feature)

- **Primary**: {KEY1} - {reason}
- **Secondary**: {KEY2} - {reason}

### Progress

{updated checklist — mark completed items [x], in-progress items as current}

**Files:**

- `{path1}` - {description}
- `{path2}` - {description}

### Notes

{blockers, decisions, findings, memories updated — or "None"}

## Previous Task

**[{OUTCOME}]**: {task_name} - {summary}

Step 4: Validate Before Writing

If ANY of these fail, gather the missing data BEFORE calling write_memory:

• Session ID matches existing WM file ([a-f0-9]{8})
• ALL daemon-managed fields preserved exactly as read
• Feature Key(s) is NOT empty or "(to be determined)"
• Task description is NOT empty or "(awaiting user task)"
• ### Progress has actual items (not placeholder text)
• **Files:** lists actual files (if any were touched)
• Status reflects reality (IN_PROGRESS / COMPLETED / BLOCKED)

Step 5: Confirm & Resume

Output: 📋 Updated Working Memory: WM_{session_id}

⚠️ CRITICAL: DO NOT STOP HERE. This skill is a utility — you MUST continue.

Exit

IMMEDIATELY resume the workflow step you were on before invoking this skill. This is a utility skill — no state change occurs. Your calling step's instructions told you to invoke /swe-wm-update as a sub-step, NOT as a stopping point.

After outputting the confirmation line above:

1. Do NOT wait for user input
2. Do NOT end your response
3. Continue with the next action from the calling WF_ step*

If you were invoked from a WF_* step's "MANDATORY NEXT STEP" table, proceed to the transition listed there. If you were invoked mid-step, continue where you left off in that step's instructions.