Save Claude Notes captures Claude Code Q&A into Markdown notes with transcript-aware defaults and flexible save targets.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install save-claude-notes
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install save-claude-notes using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Save Claude Notes?
Save Claude Notes turns the latest Claude Code conversation into a structured Markdown note, so developers can preserve the most useful questions and answers without manually copying transcript content.
Built for Openclaw Skills, it prefers --latest to read the newest substantive turn from ~/.claude/projects, then appends the saved Q&A to a daily note, an exact .md file, or a custom notes directory.
Save Claude Notes Use Cases
- Persist a solved bug, implementation decision, or architecture answer from Claude Code into a reusable knowledge base.
- Append a daily log entry to a dated Markdown note without changing your default configuration.
- Export a conversation to a specific
.mdfile for project documentation, a postmortem, or a handoff. - Save notes into a custom directory for one-off work without modifying global defaults.
- Recover when transcript discovery selects the wrong turn by overriding the project CWD or transcript path.
- Avoid repeating long answers in tool input when
--latestcan source the transcript directly.
How Save Claude Notes Works
- Receive the save command and resolve the intended target: the configured daily note, an exact file, or a custom folder.
- Prefer
--latestso the script reads the latest substantive Q&A from the Claude Code transcript instead of duplicating it in the request. - Discover transcripts from
~/.claude/projectsusing the current working directory. - If discovery is wrong or incomplete, retry in this order:
SAVE_NOTE_CWD="PROJECT_PATH" --latest,--latest --transcript PATH,--question-file PATH --answer-file PATH, then stdin JSON for short content only. - Append the Markdown entry to
<notesDir>/YYYY-MM-DD.md, an exact.mdfile, orTARGET_DIRECTORY/YYYY-MM-DD.mddepending on the selected mode. - If the user wants a permanent default location, update
~/.claude/save-claude-notes/config.jsonwith{"notesDir":"TARGET_DIRECTORY"}. - Keep the workflow manual; automatic capture requires a separate hook or wrapper.
Save Claude Notes Setup
- Ensure the skill script is available at
~/.claude/skills/save-claude-notes/scripts/save-claude-notes.sh. - Save the latest Q&A to the configured daily note:
~/.claude/skills/save-claude-notes/scripts/save-claude-notes.sh --latest
- Change the default notes directory only when requested by the user:
mkdir -p ~/.claude/save-claude-notes
cat > ~/.claude/save-claude-notes/config.json <<'JSON'
{"notesDir":"TARGET_DIRECTORY"}
JSON
- Save to a specific Markdown file without changing defaults:
~/.claude/skills/save-claude-notes/scripts/save-claude-notes.sh --latest --output-file "TARGET_DIRECTORY/notes.md"
- Save to a custom folder for a one-off daily note:
~/.claude/skills/save-claude-notes/scripts/save-claude-notes.sh --latest --notes-dir "TARGET_DIRECTORY"
Save Claude Notes Data Schema & Taxonomy
| Element | Location / Pattern | Meaning |
|---|---|---|
| Default config | ~/.claude/save-claude-notes/config.json |
Stores notesDir when customized |
| Default notes directory | ~/claude-notes |
Fallback base directory if no config exists |
| Daily note file | <notesDir>/YYYY-MM-DD.md |
Appended note file for normal saves |
| Exact file target | *.md or *.markdown path |
Appends directly to the provided Markdown file |
| Folder target | --notes-dir "PATH" |
Writes to PATH/YYYY-MM-DD.md without changing config |
| Transcript source | ~/.claude/projects |
Source tree used by --latest discovery |
| Override metadata | SAVE_NOTE_CWD, --transcript, --question-file, --answer-file, stdin JSON |
Fallback input modes when --latest fails |
- Each saved entry preserves Markdown formatting.
- Multi-line user questions remain intact after the generated title.
- The skill does not define a complex frontmatter schema; file path, date, and input source are the primary organizational metadata.
- Use one Q&A source mode per save; do not combine
--latestwith explicit question/answer inputs. - Redact sensitive context before saving, or explicitly confirm secrets handling when required.
Save Claude Notes Advanced Features
- Transcript-aware extraction with
--latest, so long answers do not need to be pasted back into the command. - CWD-sensitive transcript discovery with
SAVE_NOTE_CWDfor project-specific saves. - Multiple fallback input modes for unreliable discovery or long content: transcript path, question/answer files, or stdin JSON.
- Destination flexibility: configured daily note, exact Markdown file, or custom notes directory.
- Secret-safety handling with an explicit opt-in bypass via
SAVE_NOTE_ALLOW_SECRETS=1. - Manual operation by default, with compatibility for external hooks or wrappers when you want automation.
SKILL.md
Loading
Related Openclaw Skills
TED English Learning Assistant
Converts TED talk PDFs or pasted English transcripts into structured bilingual notes, vocabulary lists, reading questions, and Obsidian Canvas maps.
LYGO Excavationpro Music Lattice
Openclaw Skills that connect listeners and agents to the Excavationpro music portal, verify the vault and catalog, and route live discovery.
Lovense LAN Control
Openclaw Skills for Lovense LAN Control turns a zero-dependency Python CLI into safe, local-network control for Lovense toys and feedback scenes.
Agent Reputation
Openclaw Skills for Agent Reputation helps you discover, vet, register, claim, rate, and contact autonomous AI agents with provenance-aware trust and consent-first controls.
Pixel Asset Compiler
A utility to compile AI-generated pixel character reference sheets and individual actions into validated, game-ready assets and engine scenes.
PDF Label - Made In China
Batch-inserts "Made In China" into Amazon FBA label PDFs with barcode-aware positioning, test-page previews, and mixed-SKU support.
Featured*
