/inbox-check
Lightweight inbox triage across all configured email accounts. Dispatches fast subagent to fetch, then reviews with judgment. Use when user says "check my inbox", "any new emails?", "check email". See also: `draft-reply` once you pick an email to respond to.
$ npx -y skills add kbanc85/claudia --skill inbox-check --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition โ
- You can call itInvoke it directly when you want it.
- Slash command
/inbox-check
Context preview
The summary Claude sees to decide when to auto-load this skill.
Lightweight inbox triage across all configured email accounts. Dispatches fast subagent to fetch, then reviews with judgment. Use when user says "check my inbox", "any new emails?", "check email". See also: `draft-reply` once you pick an email to respond to.
SKILL.md
inbox-check.SKILL.mdname: inbox-check
description: Lightweight inbox triage across all configured email accounts. Dispatches fast subagent to fetch, then reviews with judgment. Use when user says "check my inbox", "any new emails?", "check email". See also: `draft-reply` once you pick an email to respond to.
effort-level: low
Inbox Check
Two-tier inbox triage: fetch fast, then judge what matters.
Trigger
- "Check my inbox"
- "Any new emails?"
- "Check email"
- "What's in my inbox?"
- `/inbox-check`
Architecture
This skill uses a two-tier approach to save tokens and time:
1. **Tier 1 (Haiku subagent):** Fast fetch and summarize from all available email MCP tools 2. **Tier 2 (You):** Review results with judgment, context, and relationship awareness
Process
Step 1: Detect Available Email Tools
Check which email MCP tools are available:
- Gmail tools (`gmail.*` or similar)
- Outlook tools (`outlook.*` or similar)
- Any other email integrations
If no email tools available: "I don't have access to any email accounts. Would you like to set up an email integration? Check `/connector-discovery` for options."
Step 2: Dispatch Fetch Agent
Use the Task tool to dispatch a Haiku subagent:
Subagent prompt:
"Check all available email accounts for recent messages (last 24 hours or since last check).
For each email, extract:
- From (name and email)
- Subject
- Date/time
- Brief summary (1 sentence)
- Has attachments? (yes/no)
- Seems urgent? (yes/no, based on subject and sender)
Return as a structured list, grouped by account if multiple.
Do not read full email bodies unless the subject is unclear.
Limit to 20 most recent per account."
Step 3: Review with Judgment
Once the fetch results come back, apply judgment using your knowledge of the user's relationships, priorities, and patterns:
**Categorize each email:**
| Category | Criteria | |----------|----------| | **Needs Reply** | From known contacts, asks a question, requires action | | **Worth Reading** | Relevant to active projects, from important people | | **Can Wait** | Newsletters, notifications, low-priority updates | | **Skip** | Marketing, spam that got through, irrelevant notifications |
**Use relationship context:**
- If sender is in `people/`, note the relationship context
- If related to active projects, flag the connection
- If sender has pending commitments, note those
Step 4: Present Results
**๐ฌ Inbox Check**
### Needs Reply
- **[Name]** โ [Subject] ([time ago])
[1-line context: what they need, any relationship notes]
### Worth Reading
- **[Name]** โ [Subject] ([time ago])
[Why it matters]
### Can Wait
- [count] emails (newsletters, notifications, low-priority)
[List briefly if < 5, otherwise summarize]
---
Want me to:
- Draft a reply to any of these?
- Read the full text of something?
- File any of these to memory?
Judgment Points
- Don't auto-categorize based on sender alone. Context matters.
- If unsure about urgency, err toward "Worth Reading"
- Surface anything from people the user has active commitments with
- Note if someone has emailed multiple times without a response
Tone
Quick, scannable, opinionated. The whole point is to save the user from inbox overwhelm. Be decisive about what matters and what doesn't.
Read more
name: inbox-check description: Lightweight inbox triage across all configured email accounts. Dispatches fast subagent to fetch, then reviews with judgment. Use when user says "check my inbox", "any new emails?", "check email". See also: `draft-reply` once you pick an email to respond to. effort-level: low
Inbox Check
Two-tier inbox triage: fetch fast, then judge what matters.
Trigger
- "Check my inbox"
- "Any new emails?"
- "Check email"
- "What's in my inbox?"
- `/inbox-check`
Architecture
This skill uses a two-tier approach to save tokens and time:
1. **Tier 1 (Haiku subagent):** Fast fetch and summarize from all available email MCP tools 2. **Tier 2 (You):** Review results with judgment, context, and relationship awareness
Process
Step 1: Detect Available Email Tools
Check which email MCP tools are available:
- Gmail tools (`gmail.*` or similar)
- Outlook tools (`outlook.*` or similar)
- Any other email integrations
If no email tools available: "I don't have access to any email accounts. Would you like to set up an email integration? Check `/connector-discovery` for options."
Step 2: Dispatch Fetch Agent
Use the Task tool to dispatch a Haiku subagent:
Subagent prompt: "Check all available email accounts for recent messages (last 24 hours or since last check). For each email, extract: - From (name and email) - Subject - Date/time - Brief summary (1 sentence) - Has attachments? (yes/no) - Seems urgent? (yes/no, based on subject and sender) Return as a structured list, grouped by account if multiple. Do not read full email bodies unless the subject is unclear. Limit to 20 most recent per account."
Step 3: Review with Judgment
Once the fetch results come back, apply judgment using your knowledge of the user's relationships, priorities, and patterns:
**Categorize each email:**
| Category | Criteria | |----------|----------| | **Needs Reply** | From known contacts, asks a question, requires action | | **Worth Reading** | Relevant to active projects, from important people | | **Can Wait** | Newsletters, notifications, low-priority updates | | **Skip** | Marketing, spam that got through, irrelevant notifications |
**Use relationship context:**
- If sender is in `people/`, note the relationship context
- If related to active projects, flag the connection
- If sender has pending commitments, note those
Step 4: Present Results
**๐ฌ Inbox Check** ### Needs Reply - **[Name]** โ [Subject] ([time ago]) [1-line context: what they need, any relationship notes] ### Worth Reading - **[Name]** โ [Subject] ([time ago]) [Why it matters] ### Can Wait - [count] emails (newsletters, notifications, low-priority) [List briefly if < 5, otherwise summarize] --- Want me to: - Draft a reply to any of these? - Read the full text of something? - File any of these to memory?
Judgment Points
- Don't auto-categorize based on sender alone. Context matters.
- If unsure about urgency, err toward "Worth Reading"
- Surface anything from people the user has active commitments with
- Note if someone has emailed multiple times without a response
Tone
Quick, scannable, opinionated. The whole point is to save the user from inbox overwhelm. Be decisive about what matters and what doesn't.
Terminal-based AI chief of staff. Remembers relationships, tracks commitments, helps you think strategically. Runs on Claude Code.
Repo: kbanc85/claudia
Other skills on claudia.
- /auto-research
Iteratively improve a local artifact (draft, document, page) by running a hill-climbing loop. The user names the artifact, the evaluator, and the budget. Claudia edits the artifact, scores it, keeps it if better or reverts if worse, repeats. Use when user says "iterate on this",
Open skill - /brain-monitor
Launch the Brain Monitor TUI, a real-time terminal dashboard for watching Claudia's memory system. Triggers on "brain monitor", "show dashboard", "memory dashboard", "terminal brain". See also: `brain` for a 3D graph view in the browser.
Open skill - /brain
Launch the Brain Visualizer, a real-time 3D view of memory and relationships. Triggers on "show your brain", "visualize memory", "open the brain", "memory graph". See also: `brain-monitor` for a terminal dashboard alternative.
Open skill - /build-team
Propose a personalized team of specialized agents based on the user's profile, goals, and how they actually work. Runs the proposal through an independent Checker, gates on the user's approval, and applies with rollback. Use when the user says "build my team", "set up my
Open skill - /capture-meeting
Process meeting notes or transcript to extract decisions, commitments, and insights. Use when user shares transcript or says "capture this meeting", "here are my notes from the call". See also: `meeting-prep` for pre-call briefings; `follow-up-draft` for post-meeting emails.
Open skill - /client-health
Health check across active client engagements showing status, deliverables, and concerns. Triggers on "how are my clients?", "client status", "client health check".
Open skill

