A highly interactive Telegram-exclusive utility that extracts, previews, and commits durable, semantic daily memory files from active agent threads.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install savenow
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 savenow 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 savenow?
savenow is a specialized extension designed for ecosystem setups utilizing Openclaw Skills on the Telegram interface. Its primary function is to transform fast-moving conversation transcripts into structured, durable knowledge stored directly in your workspace repository under memory/YYYY-MM-DD.md. By avoiding low-value chit-chat, temporary state logs, and redundant planning notes, it focuses exclusively on persisting structural fixes, architectural decisions, and permanent conventions.
With this tool, developers do not need to manually parse or copy-paste resolutions from their agent interactions. Instead, the agent runs a localized semantic comparison against today's existing notes, outputting clean unified diff previews with interactive Telegram inline buttons. This ensures that personal knowledge base files remain clean, consolidated, and contextually aware without accumulating repetitive duplicates.
savenow Use Cases
- Preserving root causes, configuration mappings, and stable codebase fixes immediately after an interactive debugging session.
- Documenting newly agreed-upon coding conventions or routing rules during an active chat with your coding agent.
- Consolidating fresh workspace insights into an existing memory section via smart in-place merging.
- Creating a lightweight, chronologically organized daily engineering log automatically populated by your developer-agent interactions.
How savenow Works
- Trigger Phase: The user executes
/savenowinside an active, Telegram-backed OpenClaw conversation. - Session Resolution: The skill queries the runtime session database to target the precise thread, ignoring system hooks, cron jobs, or command-only sessions.
- History Extraction & Semantic Review: The tool retrieves the recent transcript (up to 180 messages) and reads today's active memory file (
memory/YYYY-MM-DD.md). It extracts up to 5 self-contained, high-value candidate entries. - Action Assignment: Each extracted entry is semantically evaluated against existing files to decide whether it should be added (
add), combined with an existing header (merge), or ignored entirely (skip). - Staging & Interactive Preview: The candidates are saved to a temporary workspace file (
temp/savenow-entries.json) and a local utility script renders a markdown diff accompanied by Telegram inline keyboard options (✅ Applyand❌ Cancel). - Final Execution: If the user clicks
Apply, the merge script parses the JSON payload, dynamically patches the local daily markdown file, and presents a clean bulleted summary of saved items.
savenow Setup
To integrate savenow into your environment, follow these technical setup steps within your agent's workspace root:
- Ensure that Node.js is installed on your host system, as the parsing and merging workflows require executing JS files:
node --version
Verify that your OpenClaw runtime interface is bound to a Telegram bot surface with active session recording. This skill relies on the presence of
sessions_listandsessions_historytools.Deploy the two auxiliary scripts to your workspace root directory:
scripts/preview-diff.mjs: Handles candidate extraction comparison, TTL validation, and JSON generation.scripts/merge-daily-memory.mjs: Parses structural JSON and safely appends/merges content directly in Markdown targets.
- Configure write access for the agent on target paths:
temp/(for transient JSON payloads:savenow-entries.jsonandsavenow-pending.jsonwith a default 30-minute TTL)memory/(where your dailyYYYY-MM-DD.mdfiles are created and maintained)
savenow Data Schema & Taxonomy
Memory File Schema
Active notes are kept in memory/YYYY-MM-DD.md adhering to this strict Markdown pattern:
# YYYY-MM-DD
## HH:MM - Short title
- bullet 1
- bullet 2
For merged events, the tool updates the section and appends a - (merged HH:MM) tag to track the progression.
Temporary Storage Schema
During a preview lifecycle, candidate elements are stored in temp/savenow-entries.json using the following structural format:
| Attribute | Type | Description | Required |
|---|---|---|---|
candidate_index |
Integer | Order index starting at 0 | Yes |
title |
String | Extracted concise topic name | Yes |
bullets |
Array of Strings | Distinct, self-contained takeaways | Yes |
action |
String | Action choice: "add", "merge", or "skip" |
Yes |
merge_target_title |
String | Target heading in today's log (for merge action only) |
No |
reason |
String | Contextual reason for chosen action | No |
savenow Advanced Features
- Auto-Commit Interface: Run
/savenow autoto directly analyze, resolve conflicts, and commit changes instantly, bypassing preview prompts and inline keyboards entirely. - Flexible Targeting: Run
/savenow <sessionKey>to target historical session data that might sit outside the immediate chat thread. - Strict TTL Guarding: Pending state files (
temp/savenow-pending.json) are hard-coded to automatically expire after 30 minutes, preventing accidental merges from outdated developer sessions. - Semantic De-duplication Logic: Embedded mechanisms prevent write procedures if no new changes or additions are discovered, saving bandwidth and workspace clutter.
- Telegram Interface Optimization: Dynamically outputs single-row dual-button layouts with immediate callback handling for frictionless mobile workflows.
SKILL.md
Loading
Related Openclaw Skills
Strategic Marketing Director AI Agent Skill
An advanced AI brand and growth marketing director agent featuring 50+ mental models to guide projects through diagnostic, strategic, execution, and review phases.
Dream Analyzer
An uncompromising AI-driven psychological engine that bypasses comfort talk to dissect dreams and expose core personality conflicts.
Secretary Skill
An AI-powered procedural execution and goal-planning agent that solves ambiguity, overwhelm, and oblivion through automated planning pipelines.
Still Growing - Second Generation Parenting & Psychological AI Companion
An evidence-based psychological AI agent skill designed for second-generation parents to address intergenerational trauma, emotional projection, and parenting burnout.
ymailink
A highly versatile terminal email client for sending, reading, and organizing emails, complete with AI automation.
Global Market Briefing
An automated daily financial briefing tool that aggregates A-share market data and global news RSS feeds into beautifully formatted Markdown reports.
Featured*
