scratchpad
Skill1 starsChanged 5 months agoMITView on GitHub
Compaction-resilient session notepad. Save working context that survives long conversations. Three tiers — pinned (always loaded), working (session), manual (permanent).
We split this skill into the rules it gives an agent
5 rules. 0 are common, 3 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
Always: read on session start alongside MEMORY.md
- Distinctive: line 32
Use for critical session facts: "Project uses bun", "On branch feature/xyz", "Don't touch the legacy monolith"
- Distinctive: line 33
Keep SHORT — this eats into your context budget
- Very specific: line 79
When a Working note proves valuable across sessions, promote it to memory with /scratchpad --promote (moves to MEMORY.md)
- Very specific: line 80
When a session ends, stale Working entries can be pruned with /scratchpad --prune
“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Scratchpad — Compaction-Resilient Session Notes
Save important context to ~/.claude/artifacts/scratchpad.md that survives conversation compaction.
Why This Exists
During long sessions, context compaction drops earlier messages. Critical working notes — debugging breadcrumbs, temporary findings, "tried X because Y" — vanish. This gives you a persistent notepad with three priority tiers.
Different from memory: Memory is cross-session knowledge about the user/project. Scratchpad is within-session working state that might become memory later.
Commands
| Command | Action |
/scratchpad <content> | Add to Working Memory with timestamp |
/scratchpad --pin <content> | Add to Pinned Context (always loaded, 500 char limit) |
/scratchpad --save <content> | Add to Manual section (never pruned) |
/scratchpad --show | Display current scratchpad contents |
/scratchpad --prune | Remove Working entries older than 7 days |
/scratchpad --clear | Clear Working Memory (keep Pinned + Manual) |
Sections
Pinned Context (500 char limit)
- Always read on session start alongside MEMORY.md
- Use for critical session facts: "Project uses bun", "On branch feature/xyz", "Don't touch the legacy monolith"
- Keep SHORT — this eats into your context budget
- Overwrite semantics: new pin with same topic replaces old
Working Memory
- Timestamped session notes
- Auto-pruned after 7 days
- Good for: debugging breadcrumbs, temporary findings, "tried X, didn't work because Y"
- Newest entries at top
Manual
- Never auto-pruned
- User-controlled permanent notes
- Good for: team contacts, deployment commands, recurring gotchas
File Format
# Scratchpad
## Pinned
Project uses bun. On branch claude/explore-feature-xyz. Legacy module is 9K lines — restyle only, don't refactor.
## Working
- [2026-04-01 20:30] Found auth bug in UserContext — missing useEffect dependency
- [2026-04-01 19:15] pretext library installed, sidebar-layout.ts created
- [2026-04-01 18:00] 4-team audit complete, ASCII atlas at docs/ASCII-SCREEN-ATLAS.md
## Manual
- Project dev server: bun dev (port 3000)
- Team contact: team@example.com
- DB studio: bun db:studioBehavior
- Creates
~/.claude/artifacts/scratchpad.mdif it doesn't exist - Parses argument to determine section (--pin, --save, or default Working)
- Appends with timestamp for Working entries
- Warns if Pinned exceeds 500 chars
- Confirms what was saved
Integration
- Scratchpad is loaded on session start if it exists
- When a Working note proves valuable across sessions, promote it to memory with
/scratchpad --promote(moves to MEMORY.md) - When a session ends, stale Working entries can be pruned with
/scratchpad --prune
Examples
/scratchpad The sidebar width transition bug is in client-pdf.ts:800-809, re-join step
/scratchpad --pin On branch claude/explore-pretext-integration. Don't push to main.
/scratchpad --save Project uses 5-tier PDF fallback: Typst → Cloud → Puppeteer → HTML → JSON
/scratchpad --show
/scratchpad --pruneOther versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- Effect-TS/effect-smol★ 782same name
- anis-dr/skills★ 3same name
- dovoanhtruong/abap-agent-claudecode-workspace★ 3same name
- couimet/my-claude-skills★ 1same name
- opelpleple/meta-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository54 other files it gives its agents.
- AGENTS.mdAGENTS.md
- skills/deep-recon/agents.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- adversarial-researchSkill
- algorithmic-artSkill
- api-integrationSkill
- breakdownSkill
- c4-architectSkill
- canvas-designSkill
- coding-agentsSkill
- compound-engineeringSkill
- context-budgetSkill
- counterfactualSkill
- critiqueSkill
- d3js-visualizationSkill
- data-analyst-agentSkill
- database-migrationsSkill
- deep-reconSkill
- design-tasteSkill
- doc-coauthoringSkill
- e2b-sandboxSkill
- ecosystem-auditSkill
- excalidraw-diagramSkill
- executing-plansSkill
- frontend-designSkill
- fullstack-agentSkill
- grillSkill
- kickoffSkill
- mcp-builderSkill
- model-migrationSkill
- naming-the-machineSkill
- full-output-enforcementSkill
- package-researchSkill
- pdf-generationSkill
- pp2pSkill
- pp3pSkill
- prompt-engineerSkill
- prompt-groundingSkill
- redesignSkill
- requesting-code-reviewSkill
- reverse-engineer-to-learnSkill
- review-changesSkill
- shipSkill
- slack-gif-creatorSkill
- sprint-architectSkill
- techdebtSkill
- test-and-fixSkill
- threat-modelSkill
- verification-before-completionSkill
- visual-verdictSkill
- webapp-testingSkill
- web-asset-generatorSkill
- whatdidimissSkill
- writing-plansSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

