smart-compact
Skill3 starsChanged 9 days agoMITView on GitHub
Smarter /compact — shows a topic menu from the current session so you choose what survives the summary. Use when the user types /smart-compact, says 'smart compact', asks to 'compact with options', wants to 'choose what to focus the compact on', wants to reduce context or tokens while staying in control of what gets kept, or says things like 'context is getting big, help me compact smartly' or 'let's compact but I want to pick what to keep'. Surfaces the top 5 topics most-recent-first, lets the user pick by number, select all, smart-pick, custom, or default, then generates the exact /compact command to run.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- KangDongGwon/claude-skillstheir own version, first seen May 18, 2026
- s-choung/Research-Skillstheir own version, first seen May 16, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- s-choung/Research-Skills★ 36same name
- KangDongGwon/claude-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 repository36 other files it gives its agents.
- agent-statsSkill
- aristotle-deconstructorSkill
- babysit-prsSkill
- briefingSkill
- code-auditorSkill
- core-editSkill
- create-jira-ticketSkill
- create-tech-specSkill
- cyrus-tdd-engineerSkill
- doctorSkill
- dotfiles-syncSkill
- forgeSkill
- google-docsSkill
- google-driveSkill
- grill-meSkill
- handoffSkill
- lessons-reviewSkill
- mermaid-diagramsSkill
- metrics-emitSkill
- obligationsSkill
- optimus-plannerSkill
- overlay-initSkill
- performance-reviewSkill
- pr-create-from-commitsSkill
- ranger-reviewerSkill
- record-demoSkill
- review-contextSkill
- review-respondSkill
- scout-reviewerSkill
- self-evaluateSkill
- smart-statuslineSkill
- swarm-retroSkill
- team-leadSkill
- ticket-pickupSkill
- ticket-swarmSkill
- to-prdSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 51
If the user types m, show all topics (up to 10), numbered continuing from 6, then re-show the options menu so they can pick from the full …
- Distinctive: line 56
Wait for the user to respond.
- Very specific: line 62
After determining the focus, always end with this exact format so the user knows what to do — /compact is a built-in CLI command that only …
- Distinctive: line 137
If you discover something during this task that would improve this skill, propose the change and ask me to confirm before saving it.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Smart Compact
The user wants to compact the conversation but stay in control of what survives
the summary. Your job is to reflect on the conversation, surface the key topics
as a clean menu, let the user pick, then run /compact with a precise focus
string.
This costs almost nothing extra — the conversation is already in your context, so topic extraction is just reflection, not a new scan.
Step 1: Show topic menu
Show the topic menu. Extract the 5 most meaningful discussion topics from this session, sorted most recent first — the freshest work at the top since that's most likely what the user wants to preserve.
Be specific, not generic. "Discussed billing code" is bad. "OrderPaymentService null pointer bug — root cause traced to getPricingPlan()" is good.
Format:
🧠 Smart Compact — here's what we covered this session:
1. {topic — one line, specific, includes key outcome or artifact}
2. {topic}
3. {topic}
4. {topic}
5. {topic}
💬 Showing top 5 by recency — type 'm' to see all topics
What to keep in the summary?
→ Enter numbers (e.g. 1,3,5)
→ a = all of the above
→ m = show more topics
→ c = custom focus (you type it)
→ s = smart pick (I'll choose what matters most for continuing work)
→ d = default (no focus — /compact decides everything)Good topics mention: the specific thing worked on, what was found/decided/built, and any artifact (file, ticket, PR, fix). Avoid vague phrases like "discussed X" — say what the outcome was.
If the user types m, show all topics (up to 10), numbered continuing from 6, then re-show the options menu so they can pick from the full list.
Step 2: Wait for user input
Wait for the user to respond. Don't proceed until they pick.
Step 3: Confirm and compact
Based on their choice:
After determining the focus, always end with this exact format so the user knows
what to do — /compact is a built-in CLI command that only the user can run,
not Claude:
📋 Copy and run this in your chat input to compact:
/compact Focus on: {focus string}Numbers (e.g. "1,3") — combine those topics into a focused sentence:
Got it — compacting with focus on:
• {topic 1 title}
• {topic 3 title}
📋 Copy and run this in your chat input to compact:
/compact Focus on: {one sentence combining the selected topics, specific enough
that the summary will preserve the key facts, decisions, and artifacts from each}"a" (all) — cover ALL topics discussed in the session, not just the 5 shown. Reflect on the full conversation to write a comprehensive focus string:
Got it — keeping everything.
📋 Copy and run this in your chat input to compact:
/compact Focus on: {comprehensive sentence covering all topics discussed}"c" (custom) — ask them to type their focus, then:
📋 Copy and run this in your chat input to compact:
/compact Focus on: {their exact words}"d" (default) — no focus string, let /compact decide everything:
Got it — running default compact, no focus guidance.
📋 Copy and run this in your chat input to compact:
/compact"s" (smart pick) — choose the 2-3 topics most likely to matter for continuing work (prefer: decisions made, code changed, open tickets, unresolved issues over: background research, debugging dead-ends, already-resolved issues):
Smart pick — keeping the topics most useful for continuing:
• {topic A}
• {topic B}
📋 Copy and run this in your chat input to compact:
/compact Focus on: {focused sentence on chosen topics}What makes a good focus string
The focus string tells /compact what to prioritize in the summary. A good one:
- Names specific files, tickets, PRs, functions, or decisions
- Mentions what was concluded, not just what was discussed
- Is one sentence, 20-40 words
Bad: "Focus on: the work we did today" Good: "Focus on: FINOPS-4925 open PR #311134 payment validation logging, OrderPaymentService null pointer root cause at line 894, and Jira MCP auth fix via token refresh"
Maintenance
If you discover something during this task that would improve this skill, propose the change and ask me to confirm before saving it.
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.

