AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Vp Guided Focus

skill-vdustr-skills-vp-guided-focus · by VdustR

>-

No reviews yet
0 installs
38 views
0.0% view→install

Install

$ agentstack add skill-vdustr-skills-vp-guided-focus

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-vdustr-skills-vp-guided-focus)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Vp Guided Focus? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Guided Focus

Structured, one-at-a-time questioning to converge on requirements and decisions before planning or execution. Each question presents options with descriptions, trade-off differences, and a recommended choice with reasoning.

Workflow

Phase 0: Activation

  • Manual: User invokes the skill or asks to clarify first
  • Proactive: When detecting a complex/ambiguous task, ask: "This task has multiple valid approaches. Want to run guided focus first to align on direction?" — proceed only if user confirms
  • Do NOT trigger proactively for: single-file edits, bug fixes with clear reproduction steps, tasks with explicit requirements, or when the user has already described a clear plan

Phase 1: Configuration

Defaults (user can override via invocation arguments or during the session):

| Parameter | Default | Override Example | |-----------|---------|------------------| | Question count | 10 | a bare number argument (5) or --questions 5 | | Min options per question | 3 | --min-options 4 |

Phase 2: Questioning Loop

Ask questions one at a time. Wait for user response before proceeding.

Question format (use bullet list, not table — safe for narrow terminals):

### Question [current]/[total] — [Topic]

[Context sentence explaining why this question matters]

**A. [Option name]**
- Description: [what this option means]
- Difference: [how it differs from other options]
- Recommendation: Weak / Moderate / Strong

**B. [Option name]**
- Description: ...
- Difference: ...
- Recommendation: ...

**C. [Option name]**
- Description: ...
- Difference: ...
- Recommendation: ...

**Recommended: [letter] — [option name]**: [1-2 sentence reasoning]

Recommendation scale: Weak = viable but not ideal, Moderate = good fit, Strong = best fit for this context.

Question behavior rules:

  • Dependency awareness: Adapt subsequent questions based on previous answers. Skip questions that become irrelevant, add questions that become necessary. When adjusting, tell the user: "Based on your answer to Q[N], I'm [skipping Q about X / adding a question about Y] because [reason]."
  • Progress indicator: Always show [current]/[total] and update total if it changes
  • Dynamic adjustment: May increase or decrease question count based on task complexity, but never exceed 15 questions total. Always inform the user: "I'm [adding/reducing] questions because [reason]. New total: [N]."

User response flexibility:

  • Select an option (A/B/C)
  • Combine options (A+B) — if combined options conflict, explain the conflict and ask the user to choose one or describe a reconciliation
  • Provide a free-form answer
  • Ask a counter-question for clarification
  • Skip and defer: "skip" or "later" — mark as deferred, addressed in Phase 3
  • Express confidence level: "B, but not sure" → marked as uncertain in summary
  • End early: "enough", "done", or "wrap up" → immediately proceed to Phase 3

Phase 3: Deferred Question Resolution

If any questions were skipped/deferred during Phase 2:

  1. Present all deferred questions in a single list
  2. Ask once: "Want to revisit these, or proceed with them marked as undecided?"
  3. If revisit → ask each deferred question with context from other answers
  4. If proceed → mark all as undecided in summary

Phase 4: Summary

Generate a structured summary of all decisions:

## Focus Summary

**1. [Topic]**: [Decision] [confidence]
**2. [Topic]**: [Decision] [confidence]
...

### Uncertain Decisions
- **[Topic]**: [Decision marked as uncertain] — may need revisiting during implementation

### Undecided
- **[Topic]**: Deferred — [reason or context]

Confidence markers: [confirmed] | [uncertain] | [undecided]

Phase 5: Review & Handoff

After presenting the summary:

  1. Ask: "Want to modify any decisions, or add more questions?"
  2. If yes → allow edits or additional questions, regenerate summary
  3. If no → ask: "Ready to proceed? Options: plan (enter plan mode), execute (start directly)"

Guidelines

  • Don't ask obvious questions — skip questions where the answer is clearly

implied by context; never ask about facts detectable from the project

  • Reorder freely — if a user's answer makes a later question urgent,

reorder (and tell the user, as with any adjustment)

  • Wrap up early — if requirements are clear before reaching the question

count, propose to end and proceed to summary

Notes

  • Questions should progress from high-level (scope, goals, constraints) to specific (implementation details, trade-offs)
  • The first 2-3 questions should establish context and scope; detailed questions come after
  • If the user's project already has constraints (existing codebase, tech stack), detect and incorporate them rather than asking about known facts

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.