feedback-capture

Capture, document, and analyze product feedback with complete metadata and structured analysis.

Vision to Value Phase

Phase 6: Learning & Adaptation - Feedback capture is the input to the learning loop.

Prerequisites: Feedback encountered (customer, sales, support, etc.) Outputs used by: All phases (informs decisions, validates assumptions)

Auto-Initialization

Before capturing feedback, ensure the context folder structure exists. If missing:

1. Check if context/feedback/ folder exists
2. If not, create:
   • context/feedback/index.md (empty registry template)
   • context/feedback/themes.md (empty themes template)
3. When saving to context/feedback/[YYYY]/, create the year folder if needed

If the entire context structure is missing, inform the user to run /setup first.

Purpose

Every piece of feedback is valuable organizational intelligence. This skill ensures feedback is captured systematically with its source, context, and analysis - enabling pattern recognition and informed decision-making.

CRITICAL: When to Use

Agents MUST invoke /feedback-capture whenever they encounter feedback from:

• Customer conversations or quotes
• Sales call notes or deal feedback
• Support tickets or escalations
• User research findings
• Survey responses
• Product reviews or social mentions
• Internal stakeholder feedback
• Competitive win/loss information
• Partner or channel feedback

Do not let feedback pass through a conversation without capturing it.

Process

1. Extract Metadata

Gather all available context about the feedback:

Field	Description	Required
Feedback Date	When the feedback was given	Yes
Source Type	Customer / Prospect / Sales / Support / Research / Internal	Yes
Source Name	Person, company, or study name	Yes
Source Role	Title/role if known	If available
Product	Which product (for multi-product orgs, e.g., AXIA, SKYMOD)	If applicable
Feature	What feature/area the feedback relates to	Yes
Product Version	Version number if applicable	If available
Channel	How feedback was received	Yes
Customer Segment	Enterprise / SMB / Startup / etc.	If known
Contract Value	ARR or deal size if known	If available

2. Record Raw Feedback

Capture the feedback verbatim or as close to original as possible:

• Direct quotes are preferred
• If paraphrasing, note it
• Include relevant context around the quote
• Preserve the customer's language and terminology

3. Analyze the Feedback

Summary

Write a 1-2 sentence summary of the core feedback.

Key Insights

Extract 2-4 specific insights from the feedback:

• What is the underlying need or problem?
• What would success look like for this person?
• What's blocking them currently?

Sentiment Assessment

• Positive: Praise, satisfaction, advocacy
• Negative: Complaint, frustration, churn risk
• Neutral: Informational, neither positive nor negative
• Mixed: Contains both positive and negative elements

Impact Assessment

Dimension	How to Assess
Urgency	Is this blocking the customer? Time-sensitive?
Frequency	First time hearing this, or recurring pattern?
Revenue Impact	Risk to existing revenue or opportunity for expansion?
Strategic Relevance	Does this relate to an active strategic bet?

Categorization

• Type: Bug / Feature Request / Usability / Pricing / Support / General
• Topics: Assign 2-5 topic tags for searchability
• Segment: Customer segment if identifiable

4. Make Connections

Check the context registry for related items:

Related Feedback

• Run /feedback-recall [topic] to find similar past feedback
• Link to related entries if this reinforces a pattern

Linked Decisions

• Does this feedback validate or challenge a past decision?
• Reference relevant DR-IDs

Linked Bets

• Does this relate to an active strategic bet?
• Reference relevant SB-IDs

Linked Assumptions

• Does this validate or invalidate a tracked assumption?
• Reference relevant A-IDs
• If an assumption is invalidated, flag for re-decision

5. Recommend Actions

Based on the analysis:

• What should be done with this feedback?
• Who should be informed?
• Should this trigger a follow-up?

6. Save to Registry

1. Generate feedback ID: FB-[YYYY]-[NNN] (check index for next number)
2. Create full entry file: context/feedback/[YYYY]/FB-[YYYY]-[NNN].md
3. Update context/feedback/index.md with summary row
4. Check if this creates/strengthens a theme in context/feedback/themes.md

7. Report Capture

Confirm what was saved:

Feedback captured: FB-2026-015
- Source: [Customer Name] ([Segment])
- Topic: [Main topic]
- Sentiment: [Sentiment]
- Linked to: [Any connections]
- Theme contribution: [If applicable]

Output Template

# Feedback: FB-[YYYY]-[NNN]

## Metadata
| Field | Value |
|-------|-------|
| **ID** | FB-[YYYY]-[NNN] |
| **Captured Date** | [Today] |
| **Feedback Date** | [When given] |
| **Source Type** | [Type] |
| **Source Name** | [Name] |
| **Source Role** | [Role] |
| **Product** | [Product name - for multi-product orgs] |
| **Feature** | [Feature/area] |
| **Product Version** | [Version] |
| **Channel** | [Channel] |
| **Captured By** | @[agent] |

## Raw Feedback

> "[Exact quote or close paraphrase]"

[Additional context if needed]

## Analysis

### Summary
[1-2 sentence summary]

### Key Insights
1. [Insight 1]
2. [Insight 2]
3. [Insight 3]

### Sentiment
**[Positive/Negative/Neutral/Mixed]** — [Brief explanation]

### Impact Assessment
| Dimension | Rating | Notes |
|-----------|--------|-------|
| Urgency | [H/M/L] | [Why] |
| Frequency | [First/Recurring] | [Notes] |
| Revenue Impact | [H/M/L] | [Notes] |
| Strategic Relevance | [H/M/L] | [Which bet] |

### Categorization
- **Type**: [Type]
- **Topics**: [tag1], [tag2], [tag3]
- **Segment**: [Segment]

## Connections

### Related Feedback
- [FB-IDs of similar feedback]

### Linked Decisions
- [DR-IDs] — [How it relates]

### Linked Bets
- [SB-IDs] — [How it relates]

### Linked Assumptions
- [A-IDs] — [Validates/Invalidates]

## Recommended Actions
- [ ] [Action 1]
- [ ] [Action 2]

Instructions

1. When encountering ANY feedback, immediately invoke this skill
2. Ask clarifying questions if metadata is missing
3. Always capture raw feedback verbatim when possible
4. Always perform analysis - don't just store raw data
5. Always check for connections to existing context
6. Always save to the registry
7. Flag if feedback invalidates assumptions or triggers re-decisions
8. Note if feedback contributes to an emerging or established theme

Theme Detection

After saving, check if this feedback:

• Matches an existing theme → Update theme with new data point
• Shares topics with 2+ other entries → Suggest new emerging theme
• Represents a significant new pattern → Flag for theme consideration

Auto-Linking (v3)

After capturing feedback:

1. Check for ID mentions: If feedback references decisions (DR-), bets (SB-), or assumptions (A-*), create cross-reference links in context/index.json
2. Match against themes: Check if feedback content matches existing themes in context/feedback/themes.md. If so, link to theme and increment theme strength
3. Update indexes: Add to sourceIndex, sentimentIndex, and topicIndex in context/index.json

See rules/context-graph.md for the full linking specification.