Top 10 DoneClaw Skills to Automate Your Workflow

1. Email Summarizer

The email summarizer skill connects to your inbox and produces a prioritized digest of unread messages. Instead of scanning 50 emails every morning, you get a structured summary that highlights what needs your attention and what can wait.

When triggered, the agent categorizes emails into urgent (needs reply today), important (needs reply this week), informational (FYI only), and ignorable (newsletters, promotions). For each urgent and important email, it extracts the sender, subject, key ask, and suggested reply length.

This skill works best when paired with a cron schedule that runs it every morning at your preferred time. After a few weeks of use, your agent learns which senders are consistently urgent and which are consistently ignorable, improving the categorization accuracy over time through persistent memory.

# ~/.openclaw/skills/email-summarizer.md

## Trigger
/emails

## Description
Summarize unread emails with priority categorization.

## Instructions
When triggered, analyze unread messages and produce:

**Urgent** (needs reply today):
- [Sender]: [Subject] — [1-line summary of what they need]

**Important** (reply this week):
- [Sender]: [Subject] — [1-line summary]

**FYI** (no action needed):
- [count] informational emails ([brief list of senders])

**Skipped:** [count] newsletters/promotions

## Rules
- Urgent = explicit deadline today or escalation language
- Important = requires your input but no immediate deadline
- Learn from past categorization patterns in memory
- Keep total summary under 300 words

2. Daily Briefing

The daily briefing aggregates information from across your agent's memory — unread messages, upcoming tasks, recent conversation highlights, and any scheduled reminders — into a single morning report delivered via Telegram or your preferred channel.

Unlike a calendar app that only shows events, the daily briefing synthesizes information contextually. It might note that you have a meeting with a client at 2pm and remind you that in yesterday's conversation you mentioned needing to prepare a proposal for that meeting. This kind of contextual connection is only possible because of DoneClaw's persistent memory.

The briefing adapts to your routine. If you consistently ask follow-up questions about certain topics, the agent starts including those topics proactively. After a month of daily briefings, most users find the summary accurately reflects their priorities without any manual configuration.

# ~/.openclaw/skills/daily-briefing.md

## Trigger
/briefing

## Description
Generate a comprehensive morning briefing.

## Instructions
Produce a morning briefing covering:
1. **Messages**: Unread count + 1-line summary of each
2. **Today**: Calendar events and scheduled tasks
3. **Priorities**: Top 3 tasks based on deadlines and importance
4. **Reminders**: Any items you asked to be reminded about
5. **Context**: Relevant notes from recent conversations

Keep under 250 words. Deliver at scheduled time via Telegram.

## Rules
- Prioritize by deadline proximity, then importance
- Include context from memory that relates to today's tasks
- If nothing is scheduled, say so briefly — do not pad

3. File Organizer

The file organizer skill helps you maintain a structured system for documents, notes, and files that your agent manages. When you share a document or piece of information, this skill categorizes it and stores it in the appropriate memory location with consistent naming conventions.

Over time, this builds a searchable knowledge base organized by topic, date, and type. When you later ask your agent to find something — a receipt from last month, notes from a specific project, or a link you saved weeks ago — the organized structure makes retrieval fast and reliable.

The skill also runs periodic cleanup. Once a week, it reviews recent additions, suggests merges for duplicate information, and flags items that might be outdated. This prevents the common problem of persistent memory becoming a disorganized dump of everything you have ever mentioned.

# ~/.openclaw/skills/file-organizer.md

## Trigger
/organize

## Description
Categorize and store information in structured memory.

## Instructions
When the user shares information after /organize:
1. Identify the type (note, receipt, link, document, contact)
2. Extract a short title and relevant date
3. Store in memory under [type]/[YYYY-MM]/[title]
4. Confirm with: Saved: [type]/[path] — [1-line description]

When triggered with "/organize search [query]":
- Search organized memory and return matching items

When triggered with "/organize cleanup":
- Review last 30 days, flag duplicates and stale items

## Rules
- Use consistent naming: lowercase, hyphens, no spaces
- Always confirm what was saved and where
- For duplicates, suggest merging instead of auto-deleting

4. Appointment Scheduler

The appointment scheduler skill manages your calendar through natural language. Tell your agent to schedule a dentist appointment next Tuesday at 3pm, and it stores the event, sets a reminder, and can even notify you the morning of with preparation notes.

The skill tracks recurring appointments automatically. If you tell it you have a weekly team standup every Monday at 10am, it creates the recurring entry and includes it in your daily briefings. You can modify or cancel individual occurrences without affecting the series.

Integration with the daily briefing skill means your schedule is always part of your morning summary. The scheduler also detects conflicts — if you try to book something during an existing appointment, it warns you and suggests alternative times based on your typical availability patterns.

# ~/.openclaw/skills/scheduler.md

## Trigger
/schedule

## Description
Manage appointments and calendar events via natural language.

## Instructions
Parse natural language scheduling requests:
- "/schedule dentist Tuesday 3pm" → create event
- "/schedule weekly standup Monday 10am" → create recurring
- "/schedule cancel dentist Tuesday" → remove event
- "/schedule" with no args → show upcoming 7 days

Store in memory under "calendar/[YYYY-MM-DD]/[event-name]"
For recurring events, store template under "calendar/recurring/"

## Rules
- Detect and warn about time conflicts
- Set reminders: 1 day before and 1 hour before
- Support "reschedule" to move an existing event
- Include events in /briefing output when configured

5. Social Media Poster

The social media poster skill helps you draft, schedule, and track social media content. Give your agent a topic or key message, and it drafts platform-appropriate posts — shorter for Twitter/X, longer for LinkedIn, visual descriptions for Instagram.

The skill maintains a content calendar in memory. You can queue posts for specific dates and review the upcoming schedule. It also tracks which topics you have posted about recently to avoid repetition and suggests content gaps based on your posting history.

This is not an auto-posting tool — it drafts content for your review and approval. The agent learns your tone and style over time, making drafts increasingly accurate to your voice. You review, edit if needed, and post manually. The agent then logs what was published and when for calendar tracking.

# ~/.openclaw/skills/social-poster.md

## Trigger
/post

## Description
Draft and schedule social media content.

## Instructions
"/post [topic]" — Draft posts for configured platforms:
- Twitter/X: Under 280 chars, punchy, with relevant hashtags
- LinkedIn: 150-300 words, professional tone, with CTA
- Instagram: Caption with emoji, hashtag block, visual suggestion

"/post schedule [date]" — Queue the last draft for a date
"/post calendar" — Show upcoming scheduled posts
"/post history" — Show last 10 published posts

## Rules
- Never auto-post — always present draft for approval
- Learn the user's tone from past approved posts
- Avoid repeating topics from the last 2 weeks
- Suggest optimal posting times based on platform norms

6. Research Assistant

The research assistant turns your agent into a focused investigator. Give it a topic, and it produces a structured summary with key findings, conflicting viewpoints, and suggested follow-up questions. Findings are stored in memory for later reference.

What makes this skill more valuable than a Google search is context accumulation. When you research a topic you have explored before, the agent references your previous findings and builds on them. If you researched AI model pricing last month and ask about it again, the agent notes what has changed since your last research.

The skill produces structured output that is easy to scan: numbered key findings, a conflicts section for disagreements between sources, and follow-up questions that help you decide whether to dig deeper. The research is stored in memory under a searchable path, so you can reference it in future conversations.

# ~/.openclaw/skills/research.md

## Trigger
/research

## Description
Research a topic and produce a structured summary.

## Instructions
When triggered with "/research [topic]":
1. Search for current information using available tools
2. Synthesize into 3-5 key findings
3. Note conflicting or uncertain information
4. Suggest 2-3 follow-up questions
5. Store under "research/[topic-slug]" in memory

Format:
## [Topic]
**Key Findings:**
1. [finding]
2. [finding]
3. [finding]
**Conflicts:** [disagreements between sources]
**Follow-up:** [suggested next steps]

## Rules
- Reference previous research on the same topic if it exists
- Distinguish facts from opinions
- Keep total output under 500 words

7. Code Reviewer

The code reviewer skill analyzes code diffs or snippets and provides actionable feedback focused on bugs, security issues, and performance problems. It skips cosmetic nitpicks and focuses on things that could actually break in production.

Paste a diff after the trigger command, and the agent identifies potential null pointer dereferences, unhandled error cases, SQL injection vectors, missing input validation, performance bottlenecks, and race conditions. Each finding includes a specific recommendation with a code example of the fix.

Over time, the agent learns your codebase patterns and language preferences. If you consistently work in TypeScript with a specific error handling pattern, the reviewer adapts its suggestions to match your conventions. It also remembers past reviews, so if the same issue appears repeatedly, it flags it as a recurring pattern that might warrant a broader fix.

# ~/.openclaw/skills/code-reviewer.md

## Trigger
/review

## Description
Review code for bugs, security issues, and performance.

## Instructions
Analyze the provided code diff or snippet:
1. **What Changed** — 1-2 sentence summary
2. **Issues** — Bugs, security, performance problems
3. **Suggestions** — Concrete fixes with code examples
4. **Verdict** — approve / request changes / needs discussion

## Rules
- Prioritize: security > bugs > performance > style
- Provide specific line references where possible
- Skip style nitpicks unless they affect correctness
- Remember patterns from past reviews for this project

8. Expense Tracker

The expense tracker skill logs purchases and bills from natural language descriptions and maintains running totals by category and month. Tell your agent you spent $12.50 on lunch, and it logs the amount, categorizes it as food, and updates your monthly total.

At any time, you can ask for a summary — total spending this month, breakdown by category, comparison to last month, or a specific category deep dive. The agent maintains complete records in memory, so you can query historical data going back to when you started using the skill.

The skill also spots patterns. If your food spending is 30% higher than last month, it mentions it in the summary. If you have a recurring subscription that posted again, it notes it. These passive insights help you stay aware of spending trends without actively monitoring a spreadsheet.

# ~/.openclaw/skills/expense-tracker.md

## Trigger
/expense

## Description
Log expenses and track spending by category.

## Instructions
"/expense $12.50 lunch" → Log: $12.50 — food — lunch — [today]
"/expense" → Show current month summary by category
"/expense report" → Full monthly breakdown with trends
"/expense undo" → Remove last logged entry

Categories: food, transport, software, housing,
entertainment, health, utilities, other

Store under "expenses/[YYYY-MM]/[entries]" in memory.

## Rules
- Auto-categorize from description, ask if ambiguous
- Show running monthly total after each entry
- Note month-over-month changes in summaries
- Flag unusual spending patterns proactively

9. Meeting Notes

The meeting notes skill transforms raw meeting notes, transcripts, or stream-of-consciousness descriptions into structured summaries with clear action items. Paste in whatever you have — bullet points, a transcript, or even a rambling voice-to-text dump — and the agent extracts the signal.

The output follows a consistent structure: a 3-5 sentence summary, decisions made, action items with owners and deadlines, open questions, and a suggested follow-up date. This consistency makes meeting records searchable and comparable over time.

The skill stores summaries in memory with date and topic tags. When you later ask what was decided about a specific topic, the agent searches your meeting history and pulls up the relevant decisions. This is especially valuable for recurring meetings where decisions build on previous ones.

# ~/.openclaw/skills/meeting-notes.md

## Trigger
/meeting

## Description
Summarize meetings and extract action items.

## Instructions
From the provided notes or transcript, produce:
1. **Summary** — 3-5 sentences
2. **Decisions** — What was decided (bullet list)
3. **Action Items** — Who, what, by when
4. **Open Questions** — Unresolved items
5. **Follow-up** — Suggested next meeting or check-in date

Store under "meetings/[YYYY-MM-DD]-[topic-slug]" in memory.

## Rules
- Flag deadlines under 48 hours as urgent
- If no names given, use "Team" as assignee
- Keep summary factual, not interpretive
- Support "/meeting search [topic]" to query past meetings

10. Custom Webhooks

The custom webhooks skill lets you connect your DoneClaw agent to external services via HTTP webhooks. When a webhook fires, the agent processes the incoming data and takes a configured action — sending a notification, updating a record in memory, or triggering another skill.

Common webhook use cases include: receiving notifications when a GitHub PR is opened (agent summarizes the diff), getting alerted when a monitoring service detects downtime (agent checks recent changes for possible causes), or processing form submissions from a website (agent categorizes and stores the leads).

This skill turns your DoneClaw agent into a lightweight integration hub. Instead of subscribing to multiple notification services, you route everything through your agent and get a unified, intelligent processing layer. The agent can correlate events — noticing that a deployment webhook fired 10 minutes before a monitoring alert, for example — because it has the context from both events in memory.

# ~/.openclaw/skills/webhooks.md

## Trigger
/webhook

## Description
Process incoming webhooks and take configured actions.

## Instructions
"/webhook add [name] [action]" — Register a new webhook handler
"/webhook list" — Show all configured webhooks
"/webhook remove [name]" — Remove a webhook handler

When a webhook fires, the agent receives the JSON payload and:
1. Identifies which handler matches the source
2. Extracts relevant data per the handler config
3. Executes the configured action (notify, store, trigger skill)
4. Logs the event under "webhooks/[name]/[timestamp]"

Example actions:
- notify: Send a summary via Telegram
- store: Save the payload data in organized memory
- trigger: Run another skill with extracted data

## Rules
- Always log incoming webhooks for audit trail
- Validate payload structure before processing
- Rate limit: max 60 webhooks per hour per handler
- Support correlation: link related events in memory

Conclusion

These 10 skills represent the most practical automations DoneClaw users rely on daily. Each one turns a repetitive task into a single command, and the combination of persistent memory with structured skill instructions makes the output improve over time. Start with the 2-3 skills that match your biggest daily time sinks, get comfortable with how they work, and then expand your setup. The real power of DoneClaw is not any individual skill — it is having an agent that remembers your context across all of them, building a comprehensive understanding of your workflows and preferences.

Frequently asked questions