Competitive Feature Matrix Builder

Competitive Feature Matrix Builder

This skill reads a folder of competitor notes (md or txt files) and produces a comparative feature matrix with a gap analysis and recommendations. No web access or API required — fully file-based, designed for Cowork.

Input:

• Folder path containing competitor notes (one file per competitor recommended; md or txt format)
• Optional: a file named your-product.md or own-product.md for self-comparison

Output:

• Markdown response with three sections: Feature Matrix (table), Gap Analysis, and Recommendations

---

Language Detection

Detect the user's language from their message:

• If Russian (or contains Cyrillic): respond in Russian
• If English (or other Latin-script language): respond in English
• If ambiguous: respond in the language of the trigger phrase used

---

Instructions

Step 1: Validate Input

1. Check that the user provided a folder path.

• If no path given: stop and respond: "Provide the path to a folder containing competitor notes in md or txt format."

1. Scan the folder for md and txt files.

• If folder is empty or does not exist: stop and respond: "No files found at the provided path. Provide a folder path containing competitor notes in md or txt format."
• Report the number of files found: "Found N files: [list filenames]."

1. Check if any file is named your-product.md or own-product.md.

• If found: include as "Your Product" column in the matrix.
• If not found: proceed without self-comparison column; add a closing note: "Add a file named your-product.md to enable self-comparison and gap analysis."

Step 2: Extract Competitor Names and Features

1. For each file (excluding your-product.md/own-product.md if present):

• Competitor name: Use the first-level heading (# Name) if present; otherwise use the filename without extension (e.g., salesforce-notes.md → "Salesforce").
• Feature extraction: Scan the file for product features, capabilities, and attributes using these signals:
• Explicit feature lists (bullet points, numbered lists, tables)
• Sentences with action verbs: "supports", "offers", "includes", "has", "provides", "enables"
• Comparison language: "unlike X, they have…", "also includes…", "unique to them…"
• Section headings describing product areas (e.g., "Integrations", "Analytics", "Pricing")

1. For your-product.md/own-product.md (if present):

• Apply the same extraction logic.
• Label column as "Your Product" in the matrix.

1. If a file contains no extractable product features (e.g., meeting notes, financial data, contact list):

• Skip the file and log: "Skipped [filename] — no product features found."
• If all files are skipped: stop and respond: "No product features could be extracted from the provided files. Ensure your notes describe competitor product capabilities, features, or attributes."

Step 3: Normalize the Feature List

1. Compile all extracted features into a single master list.

1. Deduplicate and normalize:

• Group features that describe the same capability under different names (e.g., "REST API", "API access", "API integration" → group as "API / Integrations").
• Use the most descriptive name for the grouped row.
• Keep a footnote listing original terms grouped together (append after the matrix).

1. Sort features: group by product area if discernible (e.g., "Core Features", "Integrations", "Analytics", "Pricing/Packaging"), then alphabetically within groups.

Step 4: Build the Feature Matrix

1. Create a markdown table:

• Rows: one per normalized feature
• Columns: one per competitor + "Your Product" (if present)

1. For each cell, assign a symbol:

• ✓ — feature explicitly mentioned as present
• ✗ — feature explicitly mentioned as absent or not supported
• ? — not mentioned in the notes (unknown)

1. Add a legend line below the table:

*Legend: ✓ = confirmed present, ✗ = confirmed absent, ? = not mentioned in notes*

Edge Cases:

• If only one competitor file is present: build a single-column matrix; add note: "Gap analysis requires 2+ competitors — only the feature list is generated here."
• If files contain mostly unstructured prose (no lists or headings): extract features from paragraph context; flag: "Features extracted from unstructured text — review for accuracy before sharing."

Step 5: Gap Analysis

1. Identify gaps in Your Product (only if your-product.md was provided):

• Feature gaps = features where 2+ competitors have ✓ and Your Product has ✗ or ?
• Sort by number of competitors with the feature (descending = highest priority first)
• Assign priority: High (all competitors have it), Medium (majority), Low (minority but 2+)

1. Identify unique competitor differentiators:

• Features that only one competitor has (✓ in exactly one column) — these are niche signals to monitor, not immediate gaps

1. Identify your differentiators (only if your-product.md was provided):

• Features where Your Product has ✓ and all competitors have ✗ or ? — these are unique advantages

1. If no your-product.md: list features present in 2+ competitors as "Common features across the competitive landscape" (ordered by coverage breadth).

Step 6: Recommendations

1. Write 2–4 short recommendation bullets based on the gap analysis:

• High-priority gaps: suggest evaluating for next roadmap cycle
• Watch-list items: suggest monitoring for adoption trends
• Differentiators: suggest protecting and amplifying

1. Keep recommendations specific to feature names found — do not generate generic advice.

Step 7: Output

Assemble and output the full response in three sections: Feature Matrix → Gap Analysis → Recommendations.

Add an opening line: "Processed N files: [competitor list]. Matrix covers M features."

---

Negative Cases

• Empty folder or invalid path: Stop with message: "No files found at the provided path. Provide a folder path containing competitor notes in md or txt format."
• All files non-feature content: Stop with message: "No product features could be extracted from the provided files."
• Single file, no your-product.md: Build feature list only; note gap analysis is not possible with one competitor.

---

Output Format

Processed N files: Competitor A, Competitor B, Competitor C. Matrix covers M features.

## Feature Matrix

| Feature | Competitor A | Competitor B | Competitor C | Your Product |
|---------|:---:|:---:|:---:|:---:|
| API / Integrations | ✓ | ✓ | ✗ | ✓ |
| Mobile App | ✓ | ✗ | ✓ | ✗ |
| Analytics Dashboard | ? | ✓ | ✓ | ✗ |

*Legend: ✓ = confirmed present, ✗ = confirmed absent, ? = not mentioned in notes*

---

## Gap Analysis

**Gaps in Your Product (features 2+ competitors have that you don't):**

| Feature | Competitors With It | Priority |
|---------|---------------------|---------|
| Mobile App | Competitor A, Competitor C | High |
| Analytics Dashboard | Competitor B, Competitor C | Medium |

**Unique Competitor Differentiators (only one competitor has):**
- [Feature X]: only Competitor B — monitor for broader adoption

**Your Differentiators:**
- [Feature Y]: unique to your product — protect and amplify

---

## Recommendations

1. **High priority:** Mobile App — 2/3 competitors have it. Evaluate for next planning cycle.
2. **Medium priority:** Analytics Dashboard — 2/3 competitors have it. Gather user demand signal before committing.
3. **Watch:** [Feature X] — niche differentiator for Competitor B; track if others adopt it.

Source & license

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

• Author: KirKruglov
• Source: KirKruglov/claude-skills-kit
• License: MIT

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