deep-interview

Name: deep-interview
Author: jmstar85

151stars12forks

How to install deep-interview?

jmstar85/oh-my-githubcopilot/deep-interview

$npx -y skills add jmstar85/oh-my-githubcopilot --skill deep-interview

Installs into the current project.

Summary

Socratic deep interview with mathematical ambiguity gating. Activate when user says: deep interview, interview me, ask me everything, don't assume, make sure you understand, ouroboros, socratic, I have a vague idea, not sure exactly what I want.

Files · 1

View on GitHub

SKILL.md

1	# Deep Interview
2
3	Ouroboros-inspired Socratic questioning with mathematical ambiguity scoring. Replaces vague ideas with crystal-clear specifications by asking targeted questions that expose hidden assumptions.
4
5	## Pipeline
6	`deep-interview` → `ralplan` (consensus refinement) → `omg-autopilot` (execution)
7
8	## When to Use
9	- User has a vague idea and wants thorough requirements gathering
10	- Task is complex enough that jumping to code would waste cycles
11	- User wants mathematically-validated clarity before execution
12
13	## When NOT to Use
14	- Detailed specific request with file paths → execute directly
15	- Quick fix → delegate to @executor or `/ralph`
16	- User says "just do it" → respect their intent
17
18	## Interactive Hook Protocol
19
20	MANDATORY: Use `vscode_askQuestions` for ALL user-facing questions in this skill (when available).
21	If `vscode_askQuestions` is NOT available (e.g., Copilot CLI), present numbered options in markdown and ask the user to respond with a number or freeform text.
22	This ensures structured input collection with selectable options, consistent UX, and clear decision tracking.
23
24	### When to Fire Hooks
25	\| Trigger Point \| Question Type \| Options Required \|
26	\|---------------\|---------------\|------------------\|
27	\| Phase 2 each round \| Ambiguity-targeted question \| 3-5 options + freeform \|
28	\| Phase 3 challenges \| Assumption validation \| Yes/No + "It depends..." \|
29	\| Phase 4 spec review \| Confirm crystallized spec \| Approve / Revise / Add constraints \|
30	\| Phase 5 execution bridge \| Choose next workflow \| 5 predefined options \|
31
32	### Hook Format Rules
33	- header: Short unique ID, e.g. `"interview-round-3"`, `"spec-approval"`
34	- question: The Socratic question targeting the weakest clarity dimension
35	- options: Provide 3-5 selectable answers that represent likely user intents
36	- First option: most common/expected answer (mark as `recommended`)
37	- Last option: always include an "Other / Let me explain..." escape hatch
38	- allowFreeformInput: Always `true` — user can override any option with their own words
39	- After receiving the answer, score ambiguity immediately and report progress
40
41	### Example Hook Call
42	```
43	vscode_askQuestions([{
44	header: "goal-scope",
45	question: "What's the primary output you expect? (Ambiguity: 72% → targeting Goal Clarity)",
46	options: [
47	{ label: "A working CLI tool", recommended: true },
48	{ label: "A library/SDK for other developers" },
49	{ label: "A web application with UI" },
50	{ label: "A background service/daemon" },
51	{ label: "Other / Let me explain..." }
52	],
53	allowFreeformInput: true
54	}])
55	```
56
57	## Phases
58
59	### Phase 1: Initialize
60	1. Parse the user's idea
61	2. Detect brownfield vs greenfield (use @explore to check codebase)
62	3. For brownfield: map relevant codebase areas
63	4. Initialize ambiguity score at 100%
64
65	### Phase 2: Interview Loop
66	Repeat until ambiguity <= 20% or user exits early:
67
68	1. Generate question targeting the WEAKEST clarity dimension
69	2. HOOK: Ask ONE question via `vscode_askQuestions` with:
70	- Current ambiguity score in the question text
71	- 3-5 contextual options derived from codebase analysis and prior answers
72	- `allowFreeformInput: true` always
73	3. Score ambiguity across dimensions after receiving answer:
74	- Goal Clarity (40% weight for greenfield, 35% brownfield)
75	- Constraint Clarity (30% / 25%)
76	- Success Criteria (30% / 25%)
77	- Context Clarity (N/A / 15% for brownfield)
78	4. Report progress with dimension scores and gaps
79	5. Track ontology (key entities, stability ratio)
80
81	### Phase 3: Challenge Agents
82	- Round 4+: Contrarian - challenge core assumptions
83	- HOOK: Present assumption + counter-argument, ask user to confirm/revise via `vscode_askQuestions`
84	- Round 6+: Simplifier - probe for complexity removal
85	- HOOK: Present simplification options, ask user which can be dropped
86	- Round 8+: Ontologist - find the essence (if ambiguity > 30%)
87
88	### Phase 4: Crystallize Spec
89	When ambiguity <= threshold, generate spec to `.omg/specs/deep-interview-{slug}.md`:
90	- Goal, Constraints, Non-Goals, Acceptance Criteria
91	- Assumptions Exposed & Resolved
92	- Ontology (Key Entities) with convergence tracking
93	- Interview Transcript
94	- HOOK: Spec approval — present spec summary via `vscode_askQuestions`:
95	```
96	header: "spec-approval"
97	question: "Review the crystallized spec. Ready to proceed?"
98	options: [Approve & continue, Revise specific sections, Add more constraints, Restart interview]
99	```
100
101	### Phase 5: Execution Bridge
102	HOOK: Present execution options via `vscode_askQuestions`:
103	```
104	header: "execution-bridge"
105	question: "Spec is complete (ambi

People who liked this also viewed

6 picks

Type

deep-interview

151stars12forks

View on GitHub

How to install deep-interview?

jmstar85/oh-my-githubcopilot/deep-interview

$npx -y skills add jmstar85/oh-my-githubcopilot --skill deep-interview

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Deep Interview
2
3	Ouroboros-inspired Socratic questioning with mathematical ambiguity scoring. Replaces vague ideas with crystal-clear specifications by asking targeted questions that expose hidden assumptions.
4
5	## Pipeline
6	`deep-interview` → `ralplan` (consensus refinement) → `omg-autopilot` (execution)
7
8	## When to Use
9	- User has a vague idea and wants thorough requirements gathering
10	- Task is complex enough that jumping to code would waste cycles
11	- User wants mathematically-validated clarity before execution
12
13	## When NOT to Use
14	- Detailed specific request with file paths → execute directly
15	- Quick fix → delegate to @executor or `/ralph`
16	- User says "just do it" → respect their intent
17
18	## Interactive Hook Protocol
19
20	MANDATORY: Use `vscode_askQuestions` for ALL user-facing questions in this skill (when available).
21	If `vscode_askQuestions` is NOT available (e.g., Copilot CLI), present numbered options in markdown and ask the user to respond with a number or freeform text.
22	This ensures structured input collection with selectable options, consistent UX, and clear decision tracking.
23
24	### When to Fire Hooks
25	\| Trigger Point \| Question Type \| Options Required \|
26	\|---------------\|---------------\|------------------\|
27	\| Phase 2 each round \| Ambiguity-targeted question \| 3-5 options + freeform \|
28	\| Phase 3 challenges \| Assumption validation \| Yes/No + "It depends..." \|
29	\| Phase 4 spec review \| Confirm crystallized spec \| Approve / Revise / Add constraints \|
30	\| Phase 5 execution bridge \| Choose next workflow \| 5 predefined options \|
31
32	### Hook Format Rules
33	- header: Short unique ID, e.g. `"interview-round-3"`, `"spec-approval"`
34	- question: The Socratic question targeting the weakest clarity dimension
35	- options: Provide 3-5 selectable answers that represent likely user intents
36	- First option: most common/expected answer (mark as `recommended`)
37	- Last option: always include an "Other / Let me explain..." escape hatch
38	- allowFreeformInput: Always `true` — user can override any option with their own words
39	- After receiving the answer, score ambiguity immediately and report progress
40
41	### Example Hook Call
42	```
43	vscode_askQuestions([{
44	header: "goal-scope",
45	question: "What's the primary output you expect? (Ambiguity: 72% → targeting Goal Clarity)",
46	options: [
47	{ label: "A working CLI tool", recommended: true },
48	{ label: "A library/SDK for other developers" },
49	{ label: "A web application with UI" },
50	{ label: "A background service/daemon" },
51	{ label: "Other / Let me explain..." }
52	],
53	allowFreeformInput: true
54	}])
55	```
56
57	## Phases
58
59	### Phase 1: Initialize
60	1. Parse the user's idea
61	2. Detect brownfield vs greenfield (use @explore to check codebase)
62	3. For brownfield: map relevant codebase areas
63	4. Initialize ambiguity score at 100%
64
65	### Phase 2: Interview Loop
66	Repeat until ambiguity <= 20% or user exits early:
67
68	1. Generate question targeting the WEAKEST clarity dimension
69	2. HOOK: Ask ONE question via `vscode_askQuestions` with:
70	- Current ambiguity score in the question text
71	- 3-5 contextual options derived from codebase analysis and prior answers
72	- `allowFreeformInput: true` always
73	3. Score ambiguity across dimensions after receiving answer:
74	- Goal Clarity (40% weight for greenfield, 35% brownfield)
75	- Constraint Clarity (30% / 25%)
76	- Success Criteria (30% / 25%)
77	- Context Clarity (N/A / 15% for brownfield)
78	4. Report progress with dimension scores and gaps
79	5. Track ontology (key entities, stability ratio)
80
81	### Phase 3: Challenge Agents
82	- Round 4+: Contrarian - challenge core assumptions
83	- HOOK: Present assumption + counter-argument, ask user to confirm/revise via `vscode_askQuestions`
84	- Round 6+: Simplifier - probe for complexity removal
85	- HOOK: Present simplification options, ask user which can be dropped
86	- Round 8+: Ontologist - find the essence (if ambiguity > 30%)
87
88	### Phase 4: Crystallize Spec
89	When ambiguity <= threshold, generate spec to `.omg/specs/deep-interview-{slug}.md`:
90	- Goal, Constraints, Non-Goals, Acceptance Criteria
91	- Assumptions Exposed & Resolved
92	- Ontology (Key Entities) with convergence tracking
93	- Interview Transcript
94	- HOOK: Spec approval — present spec summary via `vscode_askQuestions`:
95	```
96	header: "spec-approval"
97	question: "Review the crystallized spec. Ready to proceed?"
98	options: [Approve & continue, Revise specific sections, Add more constraints, Restart interview]
99	```
100
101	### Phase 5: Execution Bridge
102	HOOK: Present execution options via `vscode_askQuestions`:
103	```
104	header: "execution-bridge"
105	question: "Spec is complete (ambi

People who liked this also viewed

6 picks

Type