thoth-plugin 1.2.3 → 1.2.5

package/dist/defaults/skill/morning-boot/SKILL.md

@@ -1,42 +1,231 @@
 ---
 name: morning-boot
-version: 2.0.0
-description: Operational mode (e.g. 'deep-work', 'maintenance')
+description: The master orchestrator for your morning routine. Parallelizes scans across email, calendar, and Slack, then synthesizes into a daily briefing.
 triggers: 
-inputs: 
-- name: mode
-type: markdown
-required: false
-default: auto
-output: 
+template: daily-log-template.md
+created: 2026-01-09
 updated: 2026-01-09
 ---
 
+<!--
+ARCHITECTURE REFERENCE — READ BEFORE MODIFYING
+
+This skill follows the Skill Architecture pattern documented in:
+  docs/concepts/skill-architecture.md
+
+Key concepts:
+- Skills are modular context units (workflow + template + orchestration)
+- Subagents invoke skills as context templates
+- Templates ensure consistent output quality
+- Synthesis happens in main session (not subagent) for full context access
+- Nesting enables context discovery chains
+
+DO NOT modify this skill without reading the architecture doc first.
+-->
+
 # Morning Boot Skill
 
-You are the **Daily Operations Orchestrator**. Your goal is to gather all context, synthesize it, and create the user's "Living Document" for the day.
+You are the **Daily Operations Orchestrator**. Coordinate parallel scans, synthesize results with full session context, and produce the daily briefing.
+
+---
 
-## Execution Method
+## Architecture
 
-This skill uses the **Sentinel Service** for robust, parallel execution via the SDK.
+```
+Main Session (Thoth)
+│
+├─▶ SKILL.md (this file)
+│   │
+│   ├── Phase 1: Context Discovery
+│   │
+│   ├── Phase 2: Execute OpenProse VM
+│   │   └── morning-boot.prose
+│   │       └── parallel:
+│   │           ├── email_scanner → mail-triage skill → email summary
+│   │           ├── calendar_scanner → cal-grid skill → calendar summary
+│   │           └── slack_scanner → slack-pulse skill → slack summary
+│   │
+│   ├── Phase 3: Synthesize (YOU do this, with full context)
+│   │   └── Use daily-log-template.md
+│   │
+│   ├── Phase 4: Persist outputs
+│   │
+│   └── Phase 5: Present to user
+│
+└── Back to Main Session
+```
 
-**To execute this workflow:**
+**Key insight:** Subagents do the heavy scanning (keeps main context clean). YOU do the synthesis (you have full context: spillover, projects, commitments).
+
+---
 
-1. Invoke the `trigger_workflow` tool.
-2. Workflow name: `"morning-boot"`
+## Phase 1: Context Discovery
 
-```javascript
-trigger_workflow({ workflow: "morning-boot" })
+**Step 1.1 — Get Identity:**
+
+```
+skill({ name: "context-discovery" })
 ```
 
-The Sentinel workflow handles:
-- Parallel email, calendar, and task scans
-- Synthesis of priorities
-- Creation of the daily log file
-- Generating the morning briefing
+Store:
+- `EMAIL` — User's work email
+- `KB_ROOT` — Knowledge base path
+
+**If discovery fails:** Stop and report. User needs to run `thoth link`.
+
+**Step 1.2 — Determine output path:**
+
+```
+OUTPUT_DIR = {KB_ROOT}/work/operations/daily-log/{YYYY-MM-DD}/
+```
+
+Create the directory if it doesn't exist.
+
+---
+
+## Phase 2: Execute OpenProse Workflow
+
+Load and embody the OpenProse VM. Execute `morning-boot.prose`.
+
+The .prose file spawns three scanner agents in parallel:
+- `email_scanner` → invokes `mail-triage` skill
+- `calendar_scanner` → invokes `cal-grid` skill  
+- `slack_scanner` → invokes `slack-pulse` skill
+
+Each skill:
+- Handles its own context discovery
+- Reads its own config files
+- Uses its own template for output format
+- Returns a structured summary
+
+**Receive back:** `{ email, calendar, slack }` — summaries from all three scanners.
+
+---
+
+## Phase 3: Synthesize Daily Briefing
+
+**YOU do this step** — not a subagent. You have the full session context.
+
+**Step 3.1 — Load template:**
+
+Read `daily-log-template.md` from this skill folder.
+
+**Step 3.2 — Gather additional context:**
+
+Check for spillover and commitments:
+- Read `{KB_ROOT}/work/dashboard.md` for current priorities
+- Read previous day's `daily-log.md` for unfinished items
+- Cross-reference with known projects and commitments
+
+**Step 3.3 — Apply synthesis logic:**
+
+| Input | Analysis |
+|-------|----------|
+| Email ACTION items | → Pending responses (email) |
+| Calendar meetings | → Meeting load, complexity budget |
+| Calendar focus blocks | → Deep work windows |
+| Slack RESPOND items | → Pending responses (slack) |
+| Meeting notes detected | → Handoff count for post-meeting-drill |
+| Spillover from yesterday | → Carry forward or reprioritize |
+| Dashboard priorities | → Cross-reference with today's items |
+
+**Step 3.4 — Determine Top Priority and Top 3:**
+
+Apply Executive Filter:
+1. What MUST happen today? (deadlines, blockers, commitments)
+2. What SHOULD happen today? (important but not urgent)
+3. What CAN happen today? (if time permits)
+
+**Step 3.5 — Calculate Complexity Budget:**
 
-## Technical Constraints
+| Meeting Load | Meetings | Task Capacity |
+|--------------|----------|---------------|
+| Light | 0-2 | 2-3 Heavy tasks |
+| Medium | 3-4 | 1-2 Heavy tasks |
+| Heavy | 5+ | 1 Light task max |
+
+**Step 3.6 — Fill template:**
+
+Replace placeholders in `daily-log-template.md` with synthesized values.
+
+---
+
+## Phase 4: Persist Outputs
+
+Write all files to `{OUTPUT_DIR}`:
+
+| File | Content |
+|------|---------|
+| `daily-log.md` | Synthesized briefing (from Phase 3) |
+| `mail-triage.md` | Email scan results (from email scanner) |
+| `cal-grid.md` | Calendar analysis (from calendar scanner) |
+| `slack-pulse.md` | Slack activity (from slack scanner) |
+
+---
+
+## Phase 5: Present to User
+
+Verbal summary format:
+
+```
+## Morning Briefing — {DATE}
+
+**Top Priority:** {TOP_PRIORITY}
+
+**Top 3:**
+1. {PRIORITY_1}
+2. {PRIORITY_2}
+3. {PRIORITY_3}
+
+**Day Shape:** {COMPLEXITY} day — {MEETING_COUNT} meetings, {DEEP_WORK_HOURS}h deep work available
+
+**Pending Responses:** {EMAIL_ACTION_COUNT} emails, {SLACK_RESPOND_COUNT} Slack messages
+
+**Meeting Notes:** {MEETING_NOTES_COUNT} ready for processing
+
+Files saved to: work/operations/daily-log/{DATE}/
+```
+
+---
+
+## Fallback Mode
+
+If any scanner fails:
+
+1. **Report which failed** and the error
+2. **Continue with remaining scanners** — partial data is better than none
+3. **Synthesize with available data**
+4. **Offer retry:** "Want me to retry the {failed} scan?"
+
+If ALL scanners fail:
+
+1. **Fall back to KB data:**
+   - Read `work/dashboard.md`
+   - Read `work/inbox/`
+   - Check recent daily logs
+2. **Synthesize from KB** — not as fresh, but still useful
+3. **Report the situation** to user
+
+---
+
+## Templates
+
+| Template | Purpose |
+|----------|---------|
+| `daily-log-template.md` | Final synthesized briefing |
+| `mail-triage/mail-triage-template.md` | Email scan output |
+| `cal-grid/cal-grid-template.md` | Calendar scan output |
+| `slack-pulse/slack-pulse-template.md` | Slack scan output |
+
+---
+
+## Related Files
+
+- `morning-boot.prose` — OpenProse parallel execution
+- `daily-log-template.md` — Output template
+- `work/operations/slack-map.md` — Slack channel configuration
+- `work/dashboard.md` — Current priorities (for synthesis context)
+
+---
 
-- **Do NOT** attempt to manually scan emails or calendar. Use the tool.
-- **Do NOT** use `background_task` or `prose-run`.
-- **Trust Level**: Requires Level 2+ (handled by Sentinel permissions).
+*Morning Boot Skill v3.0 | Template-driven synthesis with full context*

package/dist/defaults/skill/morning-boot/daily-log-template.md

@@ -0,0 +1,98 @@
+---
+type: operations
+hemisphere: work
+created: {{DATE}}
+updated: 2026-01-09
+tags: [daily-log, morning-boot]
+summary: Daily Briefing — {{DATE}}
+---
+
+# Daily Briefing — {{DATE}}
+
+**Generated:** {{TIME}}  
+**Scans completed:** Email ✓ | Calendar ✓ | Slack ✓
+
+---
+
+## Top Priority
+
+{{TOP_PRIORITY}}
+
+---
+
+## Top 3 for Today
+
+1. {{PRIORITY_1}}
+2. {{PRIORITY_2}}
+3. {{PRIORITY_3}}
+
+---
+
+## Complexity Budget
+
+| Metric | Value |
+|--------|-------|
+| **Meeting load** | {{MEETING_LOAD}} ({{MEETING_COUNT}} meetings, {{MEETING_HOURS}} hours) |
+| **Task capacity** | {{TASK_CAPACITY}} |
+| **Deep work available** | {{DEEP_WORK_HOURS}} hours in {{DEEP_WORK_BLOCKS}} blocks |
+
+---
+
+## Deep Work Windows
+
+| Block | Duration | Quality |
+|-------|----------|---------|
+{{DEEP_WORK_TABLE}}
+
+---
+
+## Pending Responses
+
+### Email
+
+{{EMAIL_ACTIONS}}
+
+### Slack
+
+{{SLACK_ACTIONS}}
+
+---
+
+## Calendar Overview
+
+| Time | Event | Prep | Type |
+|------|-------|------|------|
+{{CALENDAR_TABLE}}
+
+---
+
+## Meeting Notes Detected
+
+{{MEETING_NOTES_COUNT}} meeting notes ready for processing.
+
+{{#if MEETING_NOTES_COUNT > 0}}
+Run `skill({ name: "post-meeting-drill" })` to extract action items.
+
+| Meeting | Date | Message ID |
+|---------|------|------------|
+{{MEETING_NOTES_TABLE}}
+{{/if}}
+
+---
+
+## Spillover Check
+
+{{SPILLOVER_ITEMS}}
+
+---
+
+## Files Generated
+
+- `mail-triage.md` — Email scan results
+- `cal-grid.md` — Calendar analysis  
+- `slack-pulse.md` — Slack activity
+- `daily-log.md` — This file
+
+---
+
+*Generated by morning-boot skill*

package/dist/defaults/skill/morning-boot/morning-boot.prose

@@ -0,0 +1,98 @@
+# Morning Boot Workflow
+# 
+# This OpenProse program orchestrates the daily morning routine.
+# It spawns scanner agents in parallel that invoke their respective skills,
+# then synthesizes results into a daily briefing.
+#
+# Architecture:
+#   morning-boot.prose (this file)
+#   ├── email_scanner agent    → invokes skill: mail-triage
+#   ├── calendar_scanner agent → invokes skill: cal-grid
+#   └── slack_scanner agent    → invokes skill: slack-pulse
+#       ↓
+#   synthesizer agent          → combines all results
+#
+# Each skill is self-contained with its own context discovery.
+#
+# Trigger: "start my day", "morning routine", "morning boot"
+# Invoked via: skill({ name: "morning-boot" })
+
+# =============================================================================
+# AGENT DEFINITIONS
+# =============================================================================
+
+agent email_scanner:
+  model: sonnet
+  prompt: |
+    You are the Email Scanner agent for the Morning Boot workflow.
+    
+    **Your Mission**: Invoke the mail-triage skill and return its results.
+    
+    **Execution**:
+    1. Call: `skill({ name: "mail-triage" })`
+    2. Wait for the skill to complete
+    3. Return the skill's output (the mail triage report)
+    
+    The mail-triage skill handles:
+    - Context discovery (email address)
+    - Gmail API calls
+    - Classification of emails
+    - Output to `work/operations/daily-log/YYYY-MM-DD/mail-triage.md`
+    
+    **Return**: The complete mail triage report from the skill.
+
+agent calendar_scanner:
+  model: sonnet
+  prompt: |
+    You are the Calendar Scanner agent for the Morning Boot workflow.
+    
+    **Your Mission**: Invoke the cal-grid skill and return its results.
+    
+    **Execution**:
+    1. Call: `skill({ name: "cal-grid" })`
+    2. Wait for the skill to complete
+    3. Return the skill's output (the calendar grid)
+    
+    The cal-grid skill handles:
+    - Context discovery (email address)
+    - Google Calendar API calls
+    - Meeting analysis and focus block identification
+    - Output to `work/operations/daily-log/YYYY-MM-DD/cal-grid.md`
+    
+    **Return**: The complete calendar grid from the skill.
+
+agent slack_scanner:
+  model: sonnet
+  prompt: |
+    You are the Slack Scanner agent for the Morning Boot workflow.
+    
+    **Your Mission**: Invoke the slack-pulse skill and return its results.
+    
+    **Execution**:
+    1. Call: `skill({ name: "slack-pulse" })`
+    2. Wait for the skill to complete
+    3. Return the skill's output (the slack pulse report)
+    
+    The slack-pulse skill handles:
+    - Context discovery (email + kb_root)
+    - Loading slack-map.md for channel configuration
+    - Slack API calls for channels and DMs
+    - Output to `work/operations/daily-log/YYYY-MM-DD/slack-pulse.md`
+    
+    **Return**: The complete slack pulse report from the skill.
+
+# =============================================================================
+# EXECUTION FLOW
+# =============================================================================
+
+# Step 1: Parallel Skill Invocations
+# Each scanner agent invokes its respective skill
+# Skills handle their own context discovery, config loading, and templating
+parallel:
+  email = session: email_scanner
+  calendar = session: calendar_scanner
+  slack = session: slack_scanner
+
+# Step 2: Return Results to SKILL.md
+# No synthesis here — SKILL.md handles synthesis with full session context
+# This keeps the main session's context clean while subagents do heavy lifting

package/dist/defaults/skill/{_legacy/onboarding → onboarding}/SKILL.md

@@ -1,14 +1,15 @@
 ---
 name: onboarding
 description: Structured onboarding for new domains using breadth-before-depth discovery
-triggers:
-  - "Let's onboard"
-  - "New domain"
-  - "Help me set up"
-  - "Onboard my"
-  - "Learn about my"
+triggers: 
+created: 2026-01-07
+updated: 2026-01-10
 ---
 
+<!--
+ARCHITECTURE REFERENCE: docs/concepts/skill-architecture.md
+-->
+
 # Onboarding Skill
 
 You are entering **Onboarding Mode**. Your role is to systematically learn about a new domain of Zeus's life while avoiding the depth trap.