onboard

Name: onboard
Author: arezous

19stars7forks

How to install onboard?

arezous/pm-pilot/onboard

$npx -y skills add arezous/pm-pilot --skill onboard

Installs into the current project.

Summary

Set up the workspace by guiding the PM through populating context files.

Files · 1

View on GitHub

SKILL.md

1	# /onboard
2
3	Sets up the workspace by guiding the PM through populating context files.
4
5	## Trigger
6
7	- User runs `/onboard`
8	- Auto-trigger: 3 or more context files contain only a heading
9
10	## Context file structure
11
12	Route input to the right file using these structures:
13
14	- company.md — What We Do (who you are, who it's for, core problem), Market (market definition, primary JTBD, ICP segments, scope), Stage & Size, Current Priorities (top 2-3), Constraints (budget, headcount, tech debt, timeline)
15	- product.md — What It Is (value proposition), Key Differentiators (2-3 reasons to choose you), Current State (shipped, in progress, broken)
16	- competitors.md — One section per competitor: what they do, strengths, weaknesses, our angle against them
17	- personas.md — One section per persona: who they are, their goal, their pains with signal strength and a real quote
18
19	## Workflow
20
21	### 1. Read context state
22
23	Read all 4 context files (`context/company.md`, `context/competitors.md`, `context/personas.md`, `context/product.md`). Classify each as empty (only heading), thin (missing key sections), or solid (most sections filled). Don't tell the PM this classification yet. Use it to guide your questions.
24
25	### 2. Open with a braindump
26
27	Start open. Don't ask structured questions yet.
28
29	- All empty: "Tell me about your product. Dump everything: pitch deck, company description, competitor names, user research, strategy docs, a URL, whatever you've got. The messier the better, just get it out of your head. I'll sort it."
30	- Partially filled: "I see some context already. [1-2 sentences on what's solid]. What's changed recently, or what should I know that's not captured yet?"
31	- All solid: "Your context looks solid. Anything need updating, or are you here to work on something?"
32
33	If the PM seems unsure what to share: "Do you have any of these lying around? A pitch deck, company website, existing PRD, investor update, competitor list, interview notes, strategy doc. Paste, link, or drop files in `data/`. Any of these will speed things up."
34
35	### 3. Parse, tag, and write
36
37	After the PM shares, process everything in one pass:
38
39	- Route each piece to the right context file following the structure above
40	- Tag facts vs assumptions. As you write to context files, distinguish:
41	- Stated directly with evidence or specifics → write as-is
42	- Vague, aspirational, or inferred → tag with `[Assumption, verify]`
43	- Proactive web lookup: When the PM mentions their company name, offer to look it up: "Want me to search for [company] and pre-fill what I can find? You can correct anything that's wrong." If they agree, search and populate company, product, and competitors with what's publicly available. Tag all web-sourced content with `[Source: web, verify]` so the PM knows what to check.
44	- When given a URL or company name, use web search to gather real information. Don't make things up.
45	- If a URL returns little useful info, say so and ask the PM to fill the gaps manually
46	- If input touches multiple files, update all of them in one pass
47	- Never overwrite existing context. Append or refine. If new info contradicts existing info, keep both and note: `[Updated: YYYY-MM-DD] Previously X, now Y.`
48
49	### 4. Summarize back
50
51	Before asking more questions, play back what you captured. This builds trust and catches misunderstandings early.
52
53	"Here's what I got from that: [2-4 sentence summary of what you wrote]. I flagged a couple of things as assumptions. Does this sound right, or should I fix anything?"
54
55	Show the progress checklist:
56	```
57	Context status:
58	- [x] Company — solid
59	- [ ] Competitors — thin (missing strengths/weaknesses)
60	- [ ] Personas — empty
61	- [x] Product — solid
62	```
63
64	### 5. Ask pointed follow-ups
65
66	Now ask questions, but targeted, not a form. Pick 2-3 questions max based on the biggest gaps. Follow the outside-in order (company → market → competitors → personas → product) but don't force it.
67
68	Question style:
69	- Fill market gaps: "What space or market are you in?" "What problem are your customers hiring you to solve?"
70	- Probe assumptions: "You mentioned users struggle with onboarding. Is that from data, interviews, or a hunch?"
71	- Fill critical gaps: "Who are your main competitors? Even just names is fine, I can research the rest."
72	- Challenge where useful: "You said you're targeting enterprise. What makes you confident that's the right segment?"
73
74	Don't interrogate. After 2-3 questions, process the answers (step 3-4 again) before asking more. The rhythm is: braindump → summarize → 2-3 questions → summarize → 2-3 questions.
75
76	### 6. Check if context is sufficient
77
78	Minimum viable context to do useful work:
79	- Company: know what you do, who it's for, and current priorities
80	- Market: know what market you're in and the primary JTBD
81	- Product: know what exists and what's broken
82	- Competitors: at l

People who liked this also viewed

6 picks

Type

onboard

19stars7forks

View on GitHub

How to install onboard?

arezous/pm-pilot/onboard

$npx -y skills add arezous/pm-pilot --skill onboard

Installs into the current project.

Summary

Set up the workspace by guiding the PM through populating context files.

Files · 1

View on GitHub

SKILL.md

1	# /onboard
2
3	Sets up the workspace by guiding the PM through populating context files.
4
5	## Trigger
6
7	- User runs `/onboard`
8	- Auto-trigger: 3 or more context files contain only a heading
9
10	## Context file structure
11
12	Route input to the right file using these structures:
13
14	- company.md — What We Do (who you are, who it's for, core problem), Market (market definition, primary JTBD, ICP segments, scope), Stage & Size, Current Priorities (top 2-3), Constraints (budget, headcount, tech debt, timeline)
15	- product.md — What It Is (value proposition), Key Differentiators (2-3 reasons to choose you), Current State (shipped, in progress, broken)
16	- competitors.md — One section per competitor: what they do, strengths, weaknesses, our angle against them
17	- personas.md — One section per persona: who they are, their goal, their pains with signal strength and a real quote
18
19	## Workflow
20
21	### 1. Read context state
22
23	Read all 4 context files (`context/company.md`, `context/competitors.md`, `context/personas.md`, `context/product.md`). Classify each as empty (only heading), thin (missing key sections), or solid (most sections filled). Don't tell the PM this classification yet. Use it to guide your questions.
24
25	### 2. Open with a braindump
26
27	Start open. Don't ask structured questions yet.
28
29	- All empty: "Tell me about your product. Dump everything: pitch deck, company description, competitor names, user research, strategy docs, a URL, whatever you've got. The messier the better, just get it out of your head. I'll sort it."
30	- Partially filled: "I see some context already. [1-2 sentences on what's solid]. What's changed recently, or what should I know that's not captured yet?"
31	- All solid: "Your context looks solid. Anything need updating, or are you here to work on something?"
32
33	If the PM seems unsure what to share: "Do you have any of these lying around? A pitch deck, company website, existing PRD, investor update, competitor list, interview notes, strategy doc. Paste, link, or drop files in `data/`. Any of these will speed things up."
34
35	### 3. Parse, tag, and write
36
37	After the PM shares, process everything in one pass:
38
39	- Route each piece to the right context file following the structure above
40	- Tag facts vs assumptions. As you write to context files, distinguish:
41	- Stated directly with evidence or specifics → write as-is
42	- Vague, aspirational, or inferred → tag with `[Assumption, verify]`
43	- Proactive web lookup: When the PM mentions their company name, offer to look it up: "Want me to search for [company] and pre-fill what I can find? You can correct anything that's wrong." If they agree, search and populate company, product, and competitors with what's publicly available. Tag all web-sourced content with `[Source: web, verify]` so the PM knows what to check.
44	- When given a URL or company name, use web search to gather real information. Don't make things up.
45	- If a URL returns little useful info, say so and ask the PM to fill the gaps manually
46	- If input touches multiple files, update all of them in one pass
47	- Never overwrite existing context. Append or refine. If new info contradicts existing info, keep both and note: `[Updated: YYYY-MM-DD] Previously X, now Y.`
48
49	### 4. Summarize back
50
51	Before asking more questions, play back what you captured. This builds trust and catches misunderstandings early.
52
53	"Here's what I got from that: [2-4 sentence summary of what you wrote]. I flagged a couple of things as assumptions. Does this sound right, or should I fix anything?"
54
55	Show the progress checklist:
56	```
57	Context status:
58	- [x] Company — solid
59	- [ ] Competitors — thin (missing strengths/weaknesses)
60	- [ ] Personas — empty
61	- [x] Product — solid
62	```
63
64	### 5. Ask pointed follow-ups
65
66	Now ask questions, but targeted, not a form. Pick 2-3 questions max based on the biggest gaps. Follow the outside-in order (company → market → competitors → personas → product) but don't force it.
67
68	Question style:
69	- Fill market gaps: "What space or market are you in?" "What problem are your customers hiring you to solve?"
70	- Probe assumptions: "You mentioned users struggle with onboarding. Is that from data, interviews, or a hunch?"
71	- Fill critical gaps: "Who are your main competitors? Even just names is fine, I can research the rest."
72	- Challenge where useful: "You said you're targeting enterprise. What makes you confident that's the right segment?"
73
74	Don't interrogate. After 2-3 questions, process the answers (step 3-4 again) before asking more. The rhythm is: braindump → summarize → 2-3 questions → summarize → 2-3 questions.
75
76	### 6. Check if context is sufficient
77
78	Minimum viable context to do useful work:
79	- Company: know what you do, who it's for, and current priorities
80	- Market: know what market you're in and the primary JTBD
81	- Product: know what exists and what's broken
82	- Competitors: at l

People who liked this also viewed

6 picks

Type