메뉴
Persistent memory for AI coding agents via CLI. USE WHEN user says "save context", "remember this", "checkpoint", "resume session", "prepare for compaction", "wrap up", OR when starting work on a project that may span sessions.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | SaveContext-CLI |
| description | Persistent memory for AI coding agents via CLI. USE WHEN user says "save context", "remember this", "checkpoint", "resume session", "prepare for compaction", "wrap up", OR when starting work on a project that may span sessions. |
| allowed-tools | Bash(sc:*) |
Save decisions, track progress, and maintain continuity across coding sessions using the sc command-line tool.
All commands use the sc binary. Add --json for structured JSON output.
NEVER use sc session start when the user says "resume". The word "resume" (or "continue session", "pick up where I left off") ALWAYS means:
sc session list --search "<topic>" to find the sessionsc session resume <session_id> to resume itsc session start creates NEW sessions. Using it to "resume" almost always creates duplicates.
| User Says | Do This |
|---|---|
| "save this decision" | sc save "key" "value" -c decision -p high |
| "remember this" | sc save "key" "value" -c note |
| "checkpoint" | sc checkpoint create "name" |
| "wrap up" | See WrapUp workflow |
| "resume [topic]" | MUST sc session list --search "topic" then sc session resume <id> |
| "start session" | sc session start "name" -d "description" (new sessions only) |
| "what went wrong" | Check the error hint — includes suggestions, similar IDs |
Errors include hints and suggestions. If an ID is not found, check the hint for similar IDs. When no session is active, the error lists recent resumable sessions.
Structured JSON errors are automatic when piped. See cli/AGENTS.md for error codes.
sc session start "name" -d "description" # Start new session
sc session list --search "query" # Find sessions
sc session resume <id> # Resume paused session
sc session pause # Pause current session
sc session end # End session
sc status # Current session + stats
sc save "key" "value" -c decision -p high # Save context
sc get -s "search query" # Smart semantic search (auto-decomposes)
sc get -s "auth" --search-all-sessions # Search all sessions
sc get -k "exact-key" # Get by key
sc get -c decision -P high # Filter by category/priority
sc update "key" --value "new value" # Update item
sc delete "key" # Delete item
Search tips: Single keywords work best (sc get -s "auth" > sc get -s "authentication login session token handling"). Smart search auto-decomposes multi-word queries, adapts thresholds, and expands scope when no results found.
sc issue create "title" -t task -p 2 # Create issue
sc issue create "title" -t epic -p 3 # Create epic (container)
sc issue create "subtask" --parent SC-xxxx # Create subtask under epic
sc issue batch --json-input '{"issues":[...]}' # Batch with hierarchy
sc issue list -s open # List open issues
sc issue list --parent SC-xxxx # List subtasks of an epic
sc issue claim <id> # Claim for work
sc issue complete <id> # Mark complete
sc issue complete <id> --reason "Done" # Complete with reason
sc issue update <id> -s in_progress # Update status
sc issue count # Count by status
sc issue count --group-by type # Count by type/priority/assignee
sc issue stale # Stale issues (7+ days default)
sc issue stale --days 3 # Stale 3+ days
sc issue blocked # Blocked issues + blockers
sc issue dep tree <id> # Dependency tree
sc issue dep tree # Trees for all epics
sc checkpoint create "name" --include-git # Create checkpoint
sc checkpoint list -s "query" # Find checkpoints
sc checkpoint show <id> # Get details
sc checkpoint restore <id> # Restore
sc memory save "key" "value" -c command # Save memory
sc memory get "key" # Get memory
sc memory list # List all
sc plan create "title" --content "markdown" # Create plan
sc plan list # List plans
sc plan show <id> # Get plan details
sc plan capture # Import plan from AI agent's plan file
sc plan capture --agent claude --max-age 60 # Specific agent, 60min max age
sc plan capture --file /path/to/plan.md # Explicit file path
sc prime # Full context aggregation
sc prime --compact # Markdown for agent injection
sc prime --smart --compact # Relevance-ranked within token budget
sc prime --smart --compact --query "auth" # Semantic boost for topic
sc prime --smart --compact --budget 2000 # Custom token budget
sc prime --transcript # Include session transcripts
| Workflow | Trigger | File |
|---|---|---|
| QuickSave | "save", "remember", "note this" | Workflows/QuickSave.md |
| SessionStart | Starting work, "begin session" | Workflows/SessionStart.md |
| WrapUp | "wrap up", "end of day" | Workflows/WrapUp.md |
| Resume | "resume", "continue" | Workflows/Resume.md |
| Compaction | "prepare for compaction" | Workflows/Compaction.md |
| IssueTracking | "create issue", "track this bug" | Workflows/IssueTracking.md |
| Planning | "plan this feature" | Workflows/Planning.md |
| FeatureLifecycle | "implement", "build this", "plan and execute" | Workflows/FeatureLifecycle.md |
| Prime | "what's the full context", conversation start | Workflows/Prime.md |
| Advanced | Multi-day projects, multi-agent, branch switching, subagents | Workflows/AdvancedWorkflows.md |
Example 1: Quick save during work
# User: "remember that we chose JWT over sessions for auth"
sc save "auth-decision" "Chose JWT over sessions for stateless scaling" -c decision -p high
Example 2: Starting a work session
# User: "let's work on the payment feature"
sc session start "payment-feature" -d "Implementing Stripe integration"
Example 3: Wrapping up
# User: "wrap up for today"
sc save "wrapup-2025-01-30" "Completed login flow. Next: rate limiting" -c progress
sc checkpoint create "wrapup-2025-01-30" --include-git
sc session pause
Full CLI reference: Workflows/Reference.md