Agent skill

response-quality-analysis

Analyze whether your response addresses the actual question asked before posting. Use when: (1) About to post response to forum/Slack question, (2) Want to validate response coverage, (3) Need to ensure solving the right problem, (4) Want specific improvement suggestions for gaps in response

Stars 5
Forks 1

Install this agent skill to your Project

npx add-skill https://github.com/m31uk3/ai-skills/tree/main/skills/comms--response-quality-analysis

SKILL.md

Response Quality Analysis

Validates that your response actually solves the problem asked, not the problem you're comfortable addressing. Systematically analyzes problem quality, decomposes into components, calculates coverage, and provides actionable improvements.

When to Use

Before posting substantive answers to:

  • Forum questions (Slack, internal channels, Stack Overflow)
  • Mailing list responses
  • Documentation contributions
  • Any situation where you want to ensure helpfulness

Quick Start

Ask user for:

  • original_problem - The question/problem statement
  • draft_response - Your proposed answer
  • response_context - Where posting (optional)
  • work_dir - Artifact location (default: ".sop/response-analysis")

Support multiple input methods: direct paste, screenshot, link, file.

Workflow

1. Problem Intake Validation (Score: X/5)

Check original problem for required elements:

  1. ✅ Specific question (concrete, not "help me with X")
  2. ✅ Current state (what exists now)
  3. ✅ Desired state (what "solved" looks like)
  4. ⚠️ Constraints (environment, tools, culture)
  5. ❌ Prior attempts (what's been tried)

Scoring:

  • ✅ Present (clearly stated)
  • ⚠️ Partial (mentioned but vague)
  • ❌ Missing (not addressed)

Save to {work_dir}/intake-validation.md

Present to user:

Problem Quality: X/5 elements

✅ Present: [list with evidence]
⚠️ Partial: [list with what's unclear]
❌ Missing: [list with why needed]

Options:
[A] Proceed (score >= 3/5)
[B] Generate clarifying questions for OP
[C] Note limitations, proceed anyway
[D] Too vague - cannot analyze

MUST NOT proceed if < 3/5 without user confirmation

2. Problem Decomposition (N components)

Extract components ONLY from evidence in original problem.

Format each:

markdown
## Component: [Name]

**Evidence:** "[Direct quote]"
**What they're asking:** [Specific need]
**Success criteria:** [Observable outcome]
**Priority:** Critical / Important / Nice-to-have

Priority based on:

  • Critical: Core to question, must address
  • Important: Explicitly mentioned or strongly implied
  • Nice-to-have: Would add value, not strictly necessary

Target: 2-5 components (more = not distinct)

Save to {work_dir}/decomposition.md

Checkpoint:

I've identified [N] components:
1. [Name] - [Priority] - [One line]
2. [Name] - [Priority] - [One line]

Questions:
- Match your understanding?
- Missed or incorrect components?
- Adjust priorities?

MUST get user confirmation before proceeding

3. Coverage Analysis (Overall: Y%)

Map draft response to each component.

Coverage calculation:

  • 0% - Not mentioned
  • 1-40% - Mentioned but vague (no specifics)
  • 41-80% - Addressed with details but incomplete
  • 81-100% - Fully addressed, concrete, actionable

For each component:

markdown
## Component: [Name] - Coverage: X%

**Needed:** [Success criteria]

**Provided:** "[Quote from response]"

**Analysis:**
✅ Strengths: [What covered well]
❌ Gaps: [What missing/vague]
📏 Specificity: [Concrete enough?]

**Justification:** [Why this score]

Calculate weighted coverage:

  • Critical: 2x weight
  • Important: 1x weight
  • Nice-to-have: 0.5x weight

Coverage Matrix:

Component      | Priority  | Coverage | Weight | Contribution
--------------------------------------------------------
[Name]         | Critical  | X%      | 2.0    | Y
[Name]         | Important | X%      | 1.0    | Y

Save to {work_dir}/coverage-analysis.md

MUST NOT inflate scores - be honest

4. Context Grounding Check

Verify response fits original context.

Check for:

  1. Environment mismatch:

    • Different org structure?
    • Requires tools/processes not mentioned?
    • Assumes different culture?
  2. Problem substitution:

    • Solving YOUR past problem?
    • Related but different problem?
    • More general problem?
  3. Constraint violation:

    • Violates stated constraints?
    • Requires unavailable resources?
    • Assumes capabilities they lack?

Document each:

markdown
## Mismatch: [Type]
**They have:** [Their context]
**You assumed:** [Your assumption]
**Impact:** [Why matters]
**Severity:** Minor / Moderate / Critical

Summary:

Context Grounding: Strong / Moderate / Weak
- Environment: Matches / Partial / Mismatches
- Problem: Same / Related / Different  
- Constraints: Respected / Some violations / Major violations

Save to {work_dir}/context-check.md

Ask:

Are you solving THEIR problem or similar one you faced?
[A] Their exact problem
[B] Similar (need to adapt)
[C] Different (reconsider approach)

5. Response Quality Validation (Z/6 tests)

Run testable conditions:

Test 1: Restatement (Pass/Fail)

  • Can you restate problem accurately?
  • Would OP say "yes, that's it"?

Test 2: Coverage (Pass/Partial/Fail)

  • Weighted coverage >= 80%? (Pass)
  • All Critical >= 80%? (Pass)
  • = 60% OR Criticals >= 60%? (Partial)

Test 3: Specificity (Pass/Partial/Fail)

  • Can implement tomorrow?
  • Includes concrete mechanisms?

Test 4: Gap Acknowledgment (Pass/Fail)

  • Explicitly stated what not addressed?

Test 5: Context Grounding (Pass/Partial/Fail)

  • Fits their environment?
  • No critical mismatches?

Test 6: Implementation Path (Pass/Partial/Fail)

  • Can trace: current → advice → solved?
  • Steps connected without gaps?

Scoring:

  • Pass = 1.0
  • Partial = 0.5
  • Fail = 0.0

Present:

Response Quality Assessment
===========================

Coverage: X% (weighted)
Quality Score: Y/6 tests

✅ PASS: [list]
⚠️ PARTIAL: [list]
❌ FAIL: [list]

Overall: Excellent / Good / Needs Work / Poor

Critical Gap: [Most important missing piece]

Recommendation: [Specific action]

Save to {work_dir}/validation-results.md

MUST be honest - don't inflate

6. Improvement Suggestions

For each gap (coverage < 80%):

markdown
## Gap: [Component] - Currently X% coverage

**What's missing:** [Specific elements needed]

**Current text:** "[Quote or indicate missing]"

**Suggested addition:**
"[Concrete text to add - be specific enough to copy/paste]"

**Why helps:** [How addresses gap]

**Priority:** Critical / Important / Enhancement

Order by priority (Critical first)

MUST NOT provide vague suggestions: ❌ "Add more detail"
✅ "Add: 'For intake, create shared spreadsheet where Teams A,B,C submit requests with [fields]...'"

Present:

Improvement Suggestions
======================

I've identified [N] gaps:

Critical (must fix):
1. [Component]: [Gap summary]

Important (should fix):
2. [Component]: [Gap summary]

Which gaps address?
[A] All critical
[B] Critical + important
[C] Specific gaps (which?)
[D] None - proceed to decision

Save to {work_dir}/improvements.md

7. Revision Support (if requested)

Work on ONE gap at a time.

For each:

Revision: [Component]
=====================

Coverage: X% → Target: Y%

--- BEFORE ---
[Current text or "Not addressed"]

--- SUGGESTED ADDITION ---
[New text]

--- INTEGRATED RESULT ---
[How fits into full response]

New coverage: Y%

Approve?
[A] Yes, apply
[B] Modify (how?)
[C] Skip this gap

MUST:

  • Wait for approval on each
  • Update {work_dir}/draft-response.md
  • Track in {work_dir}/revision-history.md
  • Re-run coverage analysis after all revisions

Show impact:

Revision Impact
===============

Before: X% coverage, Y/6 quality
After:  X% coverage, Y/6 quality

Improved:
- [Component]: X% → Y%
- [Test]: FAIL → PASS

8. Decision Point

Present final assessment:

Final Response Assessment
=========================

Problem Quality: X/5
Response Coverage: Y%
Quality Score: Z/6

Overall: [Excellent/Good/Acceptable/Needs Work/Poor]

Strengths:
- [What covered well]

Remaining Gaps:
- [Critical/important gaps]

Context Alignment: [Strong/Adequate/Weak]

Recommendation: [Specific advice]

Options:

[A] Post as-is
    → Coverage adequate
    → Acknowledge limitations

[B] Make revisions
    → Address: [gaps]
    → Expected: X% → Y%

[C] Reconsider approach
    → May not fit context
    → Consider: [alternative]

[D] Request clarification first
    → Missing: [what need]
    → Generate questions

[E] Don't post
    → Coverage too low
    → Start over

MUST NOT recommend posting if:

  • Critical components < 50% coverage
  • Severe context mismatches
  • Quality score < 3/6

Save to {work_dir}/final-assessment.md

9. Post-Response Support (optional)

If posting, offer help with:

[A] Gap acknowledgment section:

"This addresses [covered], but doesn't cover [gaps].
For [gap], consider [pointer]."

[B] Frame limitations appropriately

  • Don't over-apologize
  • Frame as "areas for discussion"

[C] Add clarifying questions to OP

[D] Format for platform (Slack/forum)

Save final to {work_dir}/final-response.md

10. Learning Capture (optional)

Document insights in {work_dir}/learnings.md:

markdown
## Common Issue: [Pattern]
Example: "I solve feature overlap vs dependency coordination"

## Improvement
Watch for: [What to check next time]

## Meta-patterns
[Recurring blind spots]

Offer: "Analyze past responses for patterns?"

Example: Slack Response

Problem (3/5 quality): "How to handle multiple xfn dependency requests? Work flying everywhere (left diagram). Want centralized coordination (right diagram)."

Components identified:

  1. Intake mechanism (Critical) - 40% coverage
  2. Prioritization framework (Critical) - 0% coverage
  3. Communication (Important) - 30% coverage
  4. Transition path (Important) - 10% coverage

Initial coverage: 16% weighted

Critical gap: No prioritization method

After revisions: 68% coverage

Key improvement: Added concrete prioritization mechanism: "Establish weekly dependency review where TPM scores requests by impact/urgency/capacity, publishes priority queue visible to all requesters."

Troubleshooting

Low coverage despite effort:

  • Solving related but different problem
  • Go back to decomposition - reread question
  • Map each paragraph to components
  • Cut unmapped content

Context mismatches:

  • List all assumptions made
  • Check each against original
  • Use conditional language: "If you have X..."

Tests pass but feels wrong:

  • Being too generous with scores
  • Re-read original fresh
  • Ask: "Would this satisfy me?"

Takes too long:

  • Simple questions: Skip to Step 3
  • Obvious gaps: Skip validation
  • Use judgment on rigor needed

Quick Heuristics

  • Simple factual Q → Don't need full process
  • Advice/approach Q → Use full process
  • Vague/complex Q → Definitely use full process

Key Constraints

  • MUST ask for both original_problem and draft_response upfront
  • MUST NOT proceed with analysis if problem score < 3/5 without confirmation
  • MUST NOT add components without evidence in original text
  • MUST NOT inflate coverage scores - be honest
  • MUST wait for user confirmation at checkpoints
  • MUST work on one revision at a time with approval
  • MUST NOT recommend posting if critical components < 50% or quality < 3/6

Meta-Validation

Before completing, verify:

  • All 5 phases executed
  • User confirmed at checkpoints
  • Coverage honestly assessed
  • Context mismatches identified
  • Specific improvements provided
  • Final recommendation justified

Version

v1.0 - Based on universal AI workflow pattern, tested with real Slack example

Expand your agent's capabilities with these related and highly-rated skills.

m31uk3/ai-skills

guided-ooda-loop

Universal pattern for structured LLM interaction managing finite context windows through phased progression (Observe-Orient-Decide-Act). Use when the user has a complex problem, wants to design/build/create something (software, strategy, document, process), or uses phrases like "I have an idea for...", "help me design...", "guide me through...", or mentions OODA, RPI, or PDD. Reduces hallucinations through structured interaction.

5 1
Explore
m31uk3/ai-skills

doc-coauthoring

Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.

5 1
Explore
m31uk3/ai-skills

internal-comms

A set of resources to help me write all kinds of internal communications, using the formats that my company likes to use. Claude should use this skill whenever asked to write some sort of internal communications (status reports, leadership updates, 3P updates, company newsletters, FAQs, incident reports, project updates, etc.).

5 1
Explore
m31uk3/ai-skills

mcp-builder

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

5 1
Explore
m31uk3/ai-skills

canvas-design

Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.

5 1
Explore
m31uk3/ai-skills

web-artifacts-builder

Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.

5 1
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results