# Save Session

> MUST use when user says 'save', 'save session', or explicitly wants to preserve session state. {USER_NAME} does NOT use slash commands — also trigger proactively when significant work has been completed and session context should be persisted. For session endings (bye/done/goodbye), use the session-end skill instead which combines save-session + save-diary.

- **Type:** Skill
- **Install:** `agentstack add skill-faizkhairi-claude-code-blueprint-save-session`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [faizkhairi](https://agentstack.voostack.com/s/faizkhairi)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [faizkhairi](https://github.com/faizkhairi)
- **Source:** https://github.com/faizkhairi/claude-code-blueprint/tree/main/skills/save-session
- **Website:** https://github.com/faizkhairi/claude-code-blueprint#readme

## Install

```sh
agentstack add skill-faizkhairi-claude-code-blueprint-save-session
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

**Prerequisite check (run first)**: if `./memory/` does not exist OR `~/.claude/.memory-disabled` marker file is present, this skill is a no-op. Output: "Memory persistence is disabled. Run `./setup.sh` and choose to enable memory if you want this skill to work."

Save session state to memory:

1. **Update session.md** at `./memory/core/session.md`:
   - Current topic and what we're working on
   - Recent progress (what was accomplished this session)
   - Next steps (what should happen next session)
   - Any blockers or open questions

2. **Update reminders** at `./memory/core/reminders.md`:
   - Ask the user if there are any pending tasks, deadlines, or reminders to carry forward
   - Remove completed reminders (move to Completed section with date), keep active ones, add new ones

3. **Update per-project context**: If working on a registered project, update `./memory/projects/active/{project}.md` → Session Context section:
   - Last worked on (what specifically)
   - In progress (current state)
   - Next up (what should happen next on THIS project)

4. **Check preferences.md**: If new preferences or work patterns were observed during this session, update `./memory/core/preferences.md`

5. **Check MEMORY.md**: If new technical learnings or gotchas were discovered, update `{MEMORY_MD_PATH}`. After any edit, count total lines — if over 170, warn that MEMORY.md is approaching the 200-line auto-truncation limit and suggest moving content to topic files.

6. **Check decisions.md**: If a non-obvious architectural or technical decision was made this session (something future-us would ask "why?"), append to `./memory/core/decisions.md` using the existing format: `## YYYY-MM-DD — Title` / `**Context**` / `**Decision**` / `**Rationale**`

7. **Diary** (if this was a significant session — e.g., feature shipped, major bug fixed, architecture decision made): Use the `save-diary` skill to create or append an entry to `./memory/diary/current/YYYY-MM-DD.md`

8. **Git commit + push memory**: Stage all changed files in memory, commit with a descriptive message, and push to GitHub (`git push origin main`). This keeps the remote diary and session state up to date.

9. **Confirm**: Summarize what was saved and where

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [faizkhairi](https://github.com/faizkhairi)
- **Source:** [faizkhairi/claude-code-blueprint](https://github.com/faizkhairi/claude-code-blueprint)
- **License:** MIT
- **Homepage:** https://github.com/faizkhairi/claude-code-blueprint#readme

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-faizkhairi-claude-code-blueprint-save-session
- Seller: https://agentstack.voostack.com/s/faizkhairi
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.