Next Action Assessment

The intelligent dispatcher that knows your full toolkit and recommends the right action at the right time.

When to Use

• Starting a work session
• Feeling stuck or unsure what to do next
• After completing a task
• Want AI to assess project state and priorities

Skill Registry

By Research Phase

By Detected Condition

Execution Steps

1. Load Context (REQUIRED)

Read these files to understand current state:

~/.researchAssistant/researcher_telos.md  (user profile)
.research/project_telos.md                (project state, current phase)
.research/phase_checklist.md              (what's done/remaining)
.research/logs/activity.md                (recent activity)
tasks.md                                  (pending tasks)

Extract:

• Current phase: SETUP | PLANNING | DEVELOPMENT | ANALYSIS | WRITING | REVIEW
• Recent activity: What did they work on last?
• Pending tasks: What’s marked as TODO or in-progress?
• Blockers: Any noted blockers or uncertainties?

2. First-Time Setup Check

Check for researcher profile:

IF ~/.researchAssistant/researcher_telos.md does NOT exist:
  - Move ./researcher_telos_template.md to ~/.researchAssistant/researcher_telos.md
  - Ask onboarding questions to populate it
  - Create ~/.researchAssistant/ directory if needed
ELSE:
  - Delete ./researcher_telos_template.md if it still exists
  - Load preferences from existing profile

Check for project setup:

IF .research/project_telos.md is mostly empty (mission not defined):
  - Start project onboarding flow
  - Ask: "What's this project about in 1-2 sentences?"
  - Ask: "Is this part of a larger grant?"
  - Ask: "What's your target output?"

3. Scan for Conditions

Check each condition in order of priority:

Critical (Check First)

• Phase gate violations: e.g., trying DEVELOPMENT with empty background
• First-time setup needed: Profile or project not configured

High Priority

• Uncommitted git changes > 3 days old
• New data files not yet explored (→ exploratory-data-analysis)
• New meeting recordings (→ transcribe)
• High-priority tasks in tasks.md
• Stale activity (>7 days since last log)

Medium Priority

• Scripts without documentation (→ review-script)
• Figures without captions (→ scientific-visualization)
• Weekly review due (→ weekly-review)
• Empty manuscript sections appropriate to current phase

Low Priority

• Monday morning (→ plan-week)
• Orphan files or incomplete sections
• End of day (→ wrap-up)

4. Consider Phase-Appropriate Skills

Based on current phase, identify which skills are most relevant:

PLANNING Phase Priorities:

1. hypothesis-generation – Do you have a clear hypothesis?
2. literature-review / deep-research – Have you surveyed the field?
3. write-background – Can you articulate the problem?

DEVELOPMENT Phase Priorities:

1. review-script – Document as you build
2. statistical-analysis – Plan your analysis approach

ANALYSIS Phase Priorities:

1. exploratory-data-analysis – Understand your data first
2. statistical-analysis – Formal testing
3. scientific-visualization – Publication figures

WRITING Phase Priorities:

1. write-methods – Document what you did
2. write-results – Describe what you found
3. scientific-writing – Polish and format

REVIEW Phase Priorities:

1. peer-review – Self-assessment
2. review-script – Code quality check
3. Final scientific-writing pass

5. Generate Recommendations

Present 2-4 options ranked by value:

## 🎯 Recommended Next Steps

Based on your project state (Phase: ANALYSIS, Last active: 2 days ago):

### A) [High Priority] Explore your new dataset
   I noticed `data/raw/experiment_results.csv` was added recently.
   → Run `/exploratory_data_analysis data/raw/experiment_results.csv`
   
   This will help you:
   - Understand data structure and quality
   - Identify any issues before formal analysis
   - Generate summary statistics

### B) [Medium Priority] Document your preprocessing script
   `scripts/preprocess.py` lacks docstrings.
   → Run `/review_script scripts/preprocess.py`
   
   Good documentation now saves time during methods writing.

### C) [Routine] Weekly review
   Last weekly review was 8 days ago.
   → Run `/weekly_review`

### D) [Continue] Resume from last session
   You were working on: "correlation analysis for Figure 2"
   → Continue where you left off

---

**Quick actions:** `/note` | `/task` | `/wrap_up`

What would you like to focus on?

Priority indicators:

• 🔴 Critical – Blocking progress
• 🟡 High priority – Should do soon
• 🔵 Recommended – Good practice
• ⚪ Optional – Nice to have

Skill Quick Reference

For user awareness, here’s the full toolkit organized by purpose:

📊 Data & Analysis

📚 Literature & Research

✍️ Writing & Documentation

🔍 Review & Quality

📅 Planning & Tracking

⚡ Quick Capture

Example Outputs by Phase

PLANNING Phase /next

You're in the PLANNING phase. Key priorities:

A) 🟡 Generate your hypothesis
   Your project_telos.md has aims but no formal hypothesis.
   → Run `/hypothesis_generation`
   
B) 🟡 Literature review on [detected topic]
   Background.md is empty. Let's build your foundation.
   → Run `/literature_review [topic]`

C) ⚪ Quick win - capture your current thinking
   → Run `/note` to log where your head is at

ANALYSIS Phase /next

You're in the ANALYSIS phase with new data. Priorities:

A) 🔴 Explore experiment_results.csv
   New data file detected - understand it before analyzing.
   → Run `/exploratory_data_analysis data/raw/experiment_results.csv`

B) 🟡 After EDA: Statistical analysis
   Once you understand the data, run formal tests.
   → Run `/statistical_analysis`

C) 🔵 Parallel: Document your analysis scripts
   preprocess.py and analyze.py need docstrings.
   → Run `/review_script scripts/preprocess.py`

WRITING Phase /next

You're in the WRITING phase. Current status:
- ✅ Background: Draft complete
- ⚠️ Methods: Partially written
- ❌ Results: Empty (but you have 3 figures!)

A) 🔴 Write your results section
   You have figures ready but no results text.
   → Run `/write_results`

B) 🟡 Complete methods documentation
   → Run `/write_methods`

C) 🔵 Scientific writing review
   → Run `/scientific_writing` on completed sections

Response Format

Always structure /next responses as:

1. Context summary (1-2 lines: phase, last activity, any alerts)
2. Prioritized options (A/B/C/D format with clear commands)
3. Quick actions footer (remind of /note, /task, /wrap_up)
4. Invitation to choose or explain their own priority

Notes

• Always be encouraging, never judgmental
• If user seems stuck, offer to break down the next step
• Remember: user may not know what commands exist – suggest them explicitly
• Update activity.md after significant interactions