Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Merges mature lessons from a domain memory file into its instruction file. Syntax: `/memory-merger >domain [scope]` where scope is `global` (default), `user`, `workspace`, or `ws`.
.claude/skills/memory-merger/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | — | — |
| case-06 | ✗→✓ | ▲ Improved | — | — |
| case-07 | ✗→✓ | ▲ Improved | — | — |
| case-15 | ✗→✓ | ▲ Improved | — | — |
| case-21 | ✗→✓ | ▲ Improved | — | — |
You consolidate mature learnings from a domain's memory file into its instruction file, ensuring knowledge preservation with minimal redundancy.
Use the todo list to track your progress through the process steps and keep the user informed.
Memory instructions can be stored in two scopes:
global or user) - Stored in <global-prompts> (vscode-userdata:/User/prompts/) and apply to all VS Code projectsworkspace or ws) - Stored in <workspace-instructions> (<workspace-root>/.github/instructions/) and apply only to the current projectDefault scope is global.
Throughout this prompt, <global-prompts> and <workspace-instructions> refer to these directories.
/memory-merger >domain-name [scope]>domain-name - Required. The domain to merge (e.g., >clojure, >git-workflow, >prompt-engineering)[scope] - Optional. One of: global, user (both mean global), workspace, or ws. Defaults to globalExamples:
/memory-merger >prompt-engineering - merges global prompt engineering memories/memory-merger >clojure workspace - merges workspace clojure memories/memory-merger >git-workflow ws - merges workspace git-workflow memories<global-prompts>/{domain}-memory.instructions.md → <global-prompts>/{domain}.instructions.md<workspace-instructions>/{domain}-memory.instructions.md → <workspace-instructions>/{domain}.instructions.mdReview all memory sections and present them for merger consideration:
## Proposed Memories for Merger
### Memory: [Headline]
**Content:** [Key points]
**Location:** [Where it fits in instructions]
[More memories]...Say: "Please review these memories. Approve all with 'go' or specify which to skip."
STOP and wait for user input.
Establish 10/10 criteria for what constitutes awesome merged resulting instructions:
Develop the final merged instructions without updating files yet:
Once the final merged instructions meet 10/10 criteria:
applyTo patterns from both memory and instruction files if both exist, ensuring comprehensive coverage without duplicationUser: "/memory-merger >clojure"
Agent:
1. Reads clojure-memory.instructions.md and clojure.instructions.md
2. Proposes 3 memories for merger
3. [STOPS]
User: "go"
Agent:
4. Defines quality bar for 10/10
5. Merges new instructions candidate, iterates to 10/10
6. Updates clojure.instructions.md
7. Cleans clojure-memory.instructions.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +73 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.