Complete guide for Gemini Chat, Gemini CLI, and Google AI Studio.
| Tool | Best For | Setup |
|---|---|---|
| Gemini Chat (gemini.google.com) | Quick tasks, paste-and-go | None |
| Gemini CLI | Power users, local repo, best quality | Medium |
| Google AI Studio | Prototyping, system instructions, testing | Low |
| Gemini in Workspace | Google Docs/Sheets workflow integration | Low |
- Open gemini.google.com
- Upload or paste a
SKILL.mdfile - Add your request:
I've loaded a PM skill. Use it to help me [your task].
Follow the sections in order: Purpose → Key Concepts → Application.
Ask clarifying questions if needed, then generate the complete output.
Score it against the Quality Rubric at the end.
- Go to Gemini → Gem Manager → New Gem
- Name it: "PM Skills Assistant"
- Set the instructions:
You are a senior PM assistant powered by PM Skills. When the user asks for
PM help:
1. Identify which skill applies from the uploaded context.
2. Follow the skill sections in order: Purpose → Key Concepts → Application →
Examples → Pitfalls → References.
3. For Interactive skills: ask all questions BEFORE generating.
4. For Component skills: produce the full template, then self-score.
5. Target 8+/10 on all quality rubrics.
- Upload frequently used
SKILL.mdfiles - Save — every conversation with this Gem uses your skills
The Gemini CLI provides the best output quality because it reads skills directly from your filesystem.
# 1. Install Gemini CLI
npm install -g @anthropic-ai/gemini-cli
# 2. Clone the repo
git clone https://github.com/nandanosql/pm-skills.git
cd pm-skills
# 3. Start Gemini CLI
geminiSingle skill:
Read skills/strategy/okr-framework/SKILL.md and help me set Q3 OKRs
for our growth team.
Multiple skills in sequence:
First use skills/discovery/problem-statement/SKILL.md to define our problem.
Then use skills/strategy/north-star-metric/SKILL.md to establish our NSM.
Workflow skill:
Run skills/discovery/discovery-process/SKILL.md for a full discovery cycle
on our enterprise onboarding problem.
- Start from the repo root so Gemini can navigate to any skill
- Use natural language triggers — the YAML frontmatter lists trigger phrases
- Request quality scoring after each output
- Chain skills by saying "Now switch to [next skill]"
AI Studio is great for testing skills with custom system instructions and adjusting model parameters.
- Open aistudio.google.com
- Create a new Structured Prompt or Chat Prompt
- In System Instructions, paste your PM Skills operating instructions:
You are a senior PM assistant. When given a PM skill file, follow its
structure exactly:
1. Purpose — understand when and why to use this skill
2. Key Concepts — internalize the frameworks
3. Application — follow each step, use the templates
4. Examples — reference good/bad examples for calibration
5. Quality Rubric — self-score every output (target 8+/10)
6. Common Pitfalls — flag proactively
7. References — suggest related skills when relevant
- In the User turn, paste or upload a
SKILL.mdfile + your request - Adjust Temperature (recommended: 0.3–0.7 for PM outputs) and test
If your organization uses Google Workspace with Gemini enabled:
- Open a Google Doc
- Use Help me write or the Gemini side panel
- Paste the relevant skill's Application section as context
- Ask Gemini to follow the template
- For analytics skills (funnel analysis, SaaS metrics, cohort analysis)
- Paste the framework from the skill file
- Ask Gemini to help you build the spreadsheet model
Note: Workspace Gemini has shorter context windows. Upload 1 skill at a time for best results.
| Problem | Fix |
|---|---|
| Gemini doesn't follow the skill structure | Add explicit instruction: "Follow each section in order" |
| Output too brief | Say: "Follow the full Application section with all steps and templates" |
| Quality rubric ignored | Ask: "Score this against the quality rubric from the skill file" |
| File upload not available | Use copy-paste instead |
| Gem doesn't retain skills | Re-upload files and verify they appear in the Gem's knowledge |