smithery.ai

os-tk-interview

Interview-driven requirements gathering for planning

First seen Mar 22, 2026

Installation

$ npx skills add https://smithery.ai

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from smithery.ai · top by installs.

npx skills add https://smithery.ai

Browse all from smithery.ai

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 5,290 B
  • docs SUMMARY.md 75 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 1 installs

SKILL.md

/os-tk-interview "<request>" [--template <name>] [--max-questions N]

Inputs

  • Request text (from $ARGUMENTS)
  • Optional --template <template-name> to force specific template
  • Optional --max-questions N to limit questions

Resolve skill dir

SKILL_DIR=".pi/skills/os-tk-interview"
if [[ ! -d "$SKILL_DIR" ]]; then SKILL_DIR="pi/skills/os-tk-interview"; fi

Parse arguments

  • eval "$(bash "$SKILL_DIR/scripts/parse-args.sh" "$@") to populate:

- REQUEST - TEMPLATE - MAX_QUESTIONS

Behavior

  • Classify request type (refactor, feature-add, bug-fix, etc.)
  • Load appropriate interview template
  • Generate questions based on request + template
  • Ask questions and collect answers
  • Generate context summary
  • Save to dev notes (if enabled)

Workflow

Phase 1: Classify Request

Analyze the user's request to determine:

  • Request type: refactor, feature-add, bug-fix, performance, security, feature-remove
  • Complexity level: simple, medium, complex (based on keywords, scope)
  • Missing information indicators: vague language, multiple interpretations possible

Use load-template.sh script to automatically select the appropriate template.

Phase 2: Load Template

Based on request type, load interview template:

  • If --template is provided: Use that specific template
  • Otherwise: Auto-select based on request type classification

Templates are located in:

  • pi/skills/os-tk-interview/templates/refactor.md
  • pi/skills/os-tk-interview/templates/feature-add.md
  • pi/skills/os-tk-interview/templates/bug-fix.md
  • pi/skills/os-tk-interview/templates/performance.md
  • pi/skills/os-tk-interview/templates/security.md
  • pi/skills/os-tk-interview/templates/feature-remove.md

Phase 3: Generate Questions

Generate 5-10 targeted questions based on:

  • Request type (from Phase 1)
  • Template questions (from Phase 2)
  • MAX_QUESTIONS limit (if provided via --max-questions)
  • Complexity level and context

Question categories in templates:

  • Core questions: Primary intent, scope, goals
  • Technical questions: Constraints, dependencies, integration
  • Impact questions: Downstream effects, API changes, migration
  • Edge cases: Error handling, backwards compatibility

Phase 4: Collect Answers

Ask questions and collect answers. User can choose:

Interactive mode (default):

Q1: What is the primary goal?
A1: [User types answer]

Q2: What code needs to be refactored?
A2: [User types answer]

...

Batch mode (for simple requests):

I have some questions to clarify your request:

1. What is the primary goal?
2. What constraints should I consider?
3. Are there integration needs?

Please answer each question or say "N/A" if not applicable.

Answer handling:

  • Allow "N/A" or "I don't know" responses
  • Capture partial answers
  • Note gaps or areas needing more information
  • Stop when user signals "enough info" or all questions answered

Phase 5: Generate Context Summary

Create structured summary including:

# Interview Context

Request: <original request>
Type: <refactor|feature-add|bug-fix|performance|security|feature-remove>
Complexity: <simple|medium|complex>

## Answers
Q1: <question>
A1: <answer>

Q2: <question>
A2: <answer>

...

## Gaps Identified
- <any gaps still present>
- <areas needing more info>

## Next Steps
- Ready for planning with <X> questions remaining

Phase 6: Save to Dev Notes (If Enabled)

Check if dev notes are enabled in .os-tk/config.json:

  • If enabled: Save interview summary to .os-tk/dev-notes/
  • Format: Markdown with interview context
  • If there's already a dev notes for this change: Append to it
  • Otherwise: Create new timestamped note

Save using save-interview.sh script which:

  • Creates .os-tk/dev-notes/interview-<timestamp>.md or appends to change note
  • Updates .os-tk/dev-notes/INDEX.md with entry

Output Format

## Interview Complete

**Questions asked:** <number>
**Questions answered:** <number>
**Gaps remaining:** <number>

### Context Summary
[Structured summary from Phase 5]

### Ready to Proceed?

[Options]
- **"Yes, generate plan"** → Continue to planner with interview context
- **"More questions on <topic>"** → Ask follow-up questions on specific area
- **"Adjust answers"** → Review and edit answers before proceeding
- **"Save and exit"** → Save interview to dev notes without proceeding to plan

Notes

  • This is a requirements gathering step; it creates no OpenSpec change folder.
  • Use before /os-tk-plan --interview for better planning.
  • Dev notes are optional (controlled by devNotes.enabled in config).

Constraints

Question limits:

  • Default: 8-12 questions max (based on template)
  • Configurable: --max-questions N flag to override
  • Time awareness: If interview exceeds 10 minutes, ask to continue

Stop conditions:

  • User says "enough info" or "ready to plan"
  • All questions answered
  • User selects "Save and exit" option

Read-only for code:

  • Cannot modify production code, configuration, or any source files
  • Can only create/modify markdown files in .os-tk/dev-notes/
  • This is a planning step, not implementation