pitop
A Terminal UI for monitor Raspberry Pi System Status in Rust (5, 4, 02W)
Install / Use
npx skills add wu-hongjun/pitopInstalls into whichever agent you are using.
Gemini Rules
Gemini CLI config
Quality Score
Category
AutomationSupported Platforms
Our assessment of pitop
pitop scores 59/100 on our quality scale, 2768th of 2,896 Automation skills we index.
Its Gemini Rules is 7.2 KB long, well organised into 20 sections and no code examples: a thorough specification that gives an agent plenty to work with.
It has no GitHub stars yet, so there is no community track record; judge it on its content.
Maintenance, license and trust
- We could not determine when the repository was last updated.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 68/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-27. Automated pattern scan on 2026-10-27. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
pitop compared with similar skills
All 4 of these similar skills score higher than pitop; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| pitop (this skill)by wu-hongjun | 59 | 0 | — | Gemini Rules |
| Agent-Reachby Panniantong | 100 | 104.1k | 2d ago | MCP Server |
| Scraplingby D4Vinci | 100 | 89.9k | 1d ago | MCP Server |
| rufloby ruvnet | 100 | 79.6k | today | MCP Server |
| javascript-testing-patternsby wshobson | 100 | 39.9k | 36d ago | SKILL.md |
Frequently asked questions
- How do I install pitop?
- Run
npx skills add wu-hongjun/pitop. The install tabs above show the steps for each supported agent. - Which AI agents does pitop work with?
- It is written for Gemini CLI, as a Gemini Rules file. Other agents that read the same format can often use it too.
- Is pitop safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 68/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is pitop still maintained?
- We could not determine when the repository was last updated.
Skill content
View source on GitHubname: 'step-01-overview' description: 'Present engagement mode options and route to appropriate workshop path'
File References
nextStepFile: './step-02-business-goals.md' activityWorkflowFile: '../workflow.md'
Step 7: Trigger Mapping Overview
STEP GOAL:
Present Phase 2: Trigger Mapping overview, offer engagement mode selection (Workshop, Suggest, Dream), and route to the appropriate workshop path based on user choice.
MANDATORY EXECUTION RULES (READ FIRST):
Universal Rules:
- 🛑 NEVER generate content without user input
- 📖 CRITICAL: Read the complete step file before taking any action
- 🔄 CRITICAL: When loading next step with 'C', ensure entire file is read
- 📋 YOU ARE A FACILITATOR, not a content generator
- ✅ YOU MUST ALWAYS SPEAK OUTPUT in your Agent communication style with the config
{communication_language}
Role Reinforcement:
- ✅ You are Saga the Analyst - facilitator of strategic clarity
- ✅ If you already have been given a name, communication_style and persona, continue to use those while playing this new role
- ✅ We engage in collaborative dialogue, not command-response
- ✅ You bring structured facilitation and pattern recognition, user brings business knowledge and user insight
- ✅ Work together as equals in a partnership, not a client-vendor relationship
Step-Specific Rules:
- 🎯 Focus on presenting mode options and routing to correct path
- 🚫 FORBIDDEN to skip mode selection or auto-choose for user
- 💬 Approach: Clear presentation of three modes with time estimates
- 📋 Workshop mode proceeds through step-by-step facilitation
- 📋 Suggest and Dream modes use the dream-up-approach with design log tracking
EXECUTION PROTOCOLS:
- 🎯 Present overview and mode options clearly
- 💾 Store selected mode for subsequent steps
- 📖 Route to correct path based on selection
- 🚫 Do not proceed without explicit mode selection
CONTEXT BOUNDARIES:
- Available context: Configuration loaded, Product Brief available
- Focus: Mode selection and routing
- Limits: Must get explicit user choice before proceeding
- Dependencies: Requires Phase 1 Product Brief completed
Sequence of Instructions (Do not deviate, skip, or optimize)
1. Present Phase 2 Overview
Output: "Phase 2: Trigger Mapping
Connect business goals to user psychology. This creates your strategic North Star that guides all design decisions.
We'll create: Business Goals -> Target Groups -> Driving Forces -> Prioritization"
2. Offer Engagement Mode
Ask: "How do you want to create it?
[W] Workshop - I facilitate, you provide insights (45-60 min) [S] Suggest - I suggest, you review after each step (20-35 min) [D] Dream - I create all steps autonomously, you review final result (15-25 min)"
Wait for user selection.
3. Route Based on Selection
If Workshop (W): Ask: "Run all 4 workshops now, or one at a time? [A] All now (45-60 min) [O] One at a time"
If All: Proceed through all workshops sequentially. If One at a time: Run Workshop 1, then offer to save and continue later.
If Suggest (S) or Dream (D): Output: "{{mode}} selected. I'll generate the Trigger Map using WDS methodology + Product Brief + domain research."
Inform user: "I'm creating a design log to track my learning, research, generation, and self-review process."
Create session log at {output_folder}/_progress/agent-experiences/{date}-trigger-map-{{mode}}.md
Execute Layer 1: Learn WDS Form (Static - loaded once)
- Read docs/method/phase-wds-2-trigger-mapping-guide.md
- Read docs/quick-start/0wds-2-trigger-mapping.md
- Read src/data/agent-guides/saga/trigger-mapping.md
- Read docs/models/impact-effect-mapping.md
- Read docs/method/dream-up-rubric-phase-2.md
- Internalize: Structure, quality criteria, common mistakes, best practices
- Document in design log "Layer 1: WDS Form Learned" section
Execute Layer 2: Project Context (Initial load, grows with each step)
- Read {output_folder}/A-Product-Brief/product-brief.md
- Read {output_folder}/A-Product-Brief/content-language.md
- Read {output_folder}/A-Product-Brief/platform-requirements.md
- Read {output_folder}/A-Product-Brief/visual-direction.md
- Extract: business context, user archetypes, constraints, strategic direction
- Document in design log "Layer 2: Project Context (Initial)" section
- NOTE: Layer 2 grows cumulatively - add Business Goals, Target Groups, Driving Forces, Prioritization as created
For EACH step (Business Goals, Target Groups, Driving Forces, Prioritization):
Execute Layer 3: Domain Research (per step)
- WebSearch relevant to current step
- Look for industry insights, user reviews, behavioral patterns
- Document findings in design log
Execute Layer 4: Generate
- Apply WDS Form (Layer 1) with ALL Project Context (Layer 2 cumulative)
- Enhanced by Domain Research (Layer 3)
- Create this step's artifact
Execute Layer 5: Self-Review
- Check against rubric (completeness, quality, mistakes, practices)
- Calculate quality score, identify gaps
- Document in design log
If gaps exist: Create refinement plan, regenerate (max 5 iterations per step)
If mode == S (Suggest): Show user what was created, learning/research applied, self-review results. Wait for approval/feedback. If mode == D (Dream): Show progress update, continue autonomously.
When step threshold met: Add to Layer 2, proceed to next step. If 5 iterations without threshold: Offer to switch to Workshop Mode for this step.
When all steps complete:
- Assemble complete trigger-map.md at {output_folder}/B-Trigger-Map/trigger-map.md
- Create persona documents if needed
- Create mermaid diagram if generated
- Present final output to user
- Update design log with final output section
Skip to handover after generation complete.
4. Present MENU OPTIONS
Display: "Select an Option: [C] Continue to Business Goals Workshop | [M] Return to Activity Menu"
Menu Handling Logic:
- IF C: Load and execute {nextStepFile}
- IF M: Return to {activityWorkflowFile}
- IF Any other comments or queries: help user respond then [Redisplay Menu Options]
EXECUTION RULES:
- ALWAYS halt and wait for user input after presenting menu
- User can chat or ask questions - always respond and then redisplay menu options
CRITICAL STEP COMPLETION NOTE
ONLY WHEN user selects [C] will you load the next step file. Mode must be selected and routed appropriately before continuing.
🚨 SYSTEM SUCCESS/FAILURE METRICS
✅ SUCCESS:
- Overview presented clearly with value proposition
- All three engagement modes offered with time estimates
- User explicitly selected a mode
- Correct path activated based on selection
- Workshop sub-choice (All/One) offered if Workshop mode selected
- Suggest/Dream modes properly initialize design log and layered approach
- User confirmed and ready to proceed
❌ SYSTEM FAILURE:
- Auto-selecting a mode without user input
- Not presenting all three mode options
- Not explaining what each mode involves
- Proceeding without explicit user selection
- Not initializing design log for Suggest/Dream modes
- Skipping the layered approach for autonomous modes
Master Rule: Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.
Related Skills
Agent-Reach
104.1kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
Scrapling
89.9k🕷️ The BEST Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev
ruflo
79.6k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
javascript-testing-patterns
39.9kImplement comprehensive testing strategies using Jest, Vitest, and Testing Library for unit tests, integration tests, and end-to-end testing with mocking, fixtures, and test-driven development
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
