update-memory-bank

name	update-memory-bank
description	Update memory bank after task completion. Detects scope automatically and updates files in priority order. Can be auto-invoked when task completion is detected in conversation.
argument-hint	["task-description"]
disable-model-invocation	false
user-invocable	true

Update Memory Bank

Update the memory bank to reflect: $ARGUMENTS

Automatic Trigger Conditions

This skill should be invoked automatically when any of these signals are detected:

• Task implementation completed (code changes committed or ready)
• Sprint task marked complete
• Bug fix implemented
• Feature implementation finished
• Architectural decision made
• User says: "done", "finished", "complete", "task completed", "what's next?"

Process

Step 1 — Detect scope

Determine the update scope based on what changed:

Scope	When	Files to Update
Minimal	Single task completed	sprint-tracker + active-context
Standard	Feature implemented	+ progress
Full	Sprint completed / major milestone	+ archive + arch/tech files
Architecture	Architecture decision made	+ arch/patterns-overview + arch/decisions-summary
Tech Stack	New dependency or technology added	+ relevant tech/*.md

Step 2 — Update files in priority order

Priority 1 — core/sprint-tracker.md

• Mark task as complete with actual hours
• Update sprint completion percentage
• Note any scope changes

Priority 2 — core/active-context.md

• Remove completed items from "current work"
• Add to "last completed" list (keep only 5 most recent)
• Update "next steps" to reflect current state
• Update any resolved blockers

Priority 3 — core/progress.md (standard/full scope only)

• Update "what's working" section
• Update key metrics if changed
• Update overall status

Priority 4 — tech/*.md (if tech stack changed)

• New library added → update tech stack file
• New pattern discovered → add to patterns section

Priority 5 — arch/*.md (if architecture changed)

• New ADR → add to arch/decisions-summary.md
• New pattern → add to arch/patterns-overview.md

Priority 6 — Archive (full scope / sprint boundary only)

• Move completed sprint content from sprint-tracker to progress-archive.md
• Trim progress.md to summary only

Step 3 — Update dynamic zone

Update the dynamic section of project-rules.md (below === DYNAMIC CONTENT ===):

• Current sprint context
• Recent patterns catalog (add if new pattern discovered, keep last 5-10)
• Last memory bank update metadata

Step 4 — Report

Memory Bank Updated

Scope: [minimal/standard/full/architecture/tech-stack]
Trigger: [task description]

Updated:
- core/sprint-tracker.md → [what changed]
- core/active-context.md → [what changed]
- [other files if applicable]

Next steps:
- [what's next based on active-context]

File Size Check

After updating, check file sizes against limits:

File	Max Lines	Action if Exceeded
active-context.md	200	Archive to progress-archive.md
sprint-tracker.md	600	Archive sprint section to progress-archive.md
progress.md	600	Archive to progress-archive.md
tech/*.md	600	Split into sub-domain files
arch/patterns-overview.md	350	Move details to patterns-detailed.md

Rules

• Do NOT ask permission — auto-update when completion is detected
• Be smart about scope — don't update files unrelated to the completed task
• Keep reports concise — bullet points, not paragraphs
• Continue workflow — after reporting, suggest the next task
• Never skip active-context.md — it's the most critical file for session continuity