Competitive Intel Mining

competitive-intel-mining

Purpose — Find seed users by mining competitor products' public communities (GitHub issues, Discord, Reddit, forums) for people currently hitting the pain your product solves.

Triggers — "find seed users from ", "mine 's community", "who complains about on "

Why this beats buying a list — People in competitor communities have already (a) paid the cost of trying a product in your category, (b) encountered real pain with it, and (c) publicly written about the pain. That's the highest-intent signal short of a referral.

---

Inputs

• Target competitor(s) — product name + 1–3 public community URLs (GitHub repo, Discord invite, subreddit, forum)
• Pain keywords — 3–8 phrases that describe the problem your product solves (e.g., "agent loop", "sandbox rejects", "can't hold project memory")
• Pool size target — how many qualified leads you need (default: 20–80)

Outputs

• A CSV with columns: handle, platform, source_url, exact_quote, pain_category, intent_score, last_active, contact_path
• A short seeding-rationale.md summarizing: top 3 pain clusters, which competitor is ripest, recommended outreach angle

Steps

1. Enumerate sources — List every public community the competitor owns. GitHub issues/discussions are usually richest; Discord and Reddit next.
2. Scrape recent 90 days — Pull titles + bodies + reactions. For GitHub, gh api or the Issues API. For Reddit, pushshift or the JSON endpoints. For Discord, manual scroll.
3. Keyword filter — Match each item against pain keywords. Require ≥1 hit in title OR body.
4. Score intent (0–3):

• 3 — Author complaining about active blocker, asked for fix, no satisfactory answer
• 2 — Author confirmed the pain but moved on or self-patched
• 1 — Passerby commented on someone else's complaint
• 0 — Just mentioned the keyword, not actually pained

1. Deduplicate — Same handle across multiple issues → keep highest-intent row.
2. Enrich contact path — For each handle, find the easiest reply channel: X profile, GitHub bio email, personal site. Prefer the channel they're most active on.
3. Output CSV + rationale — Ship both files to case-studies/-seed-pool.csv (redacted as needed).

Quality bar

• [ ] Every row has an exact_quote — if you can't cite their words, they don't belong in the pool
• [ ] At least 60% of rows scored 2 or 3
• [ ] seeding-rationale.md answers: which competitor is ripest and why, which pain cluster to lead with
• [ ] No row depends on a private channel you can't actually reach

Anti-patterns

• Don't scrape handles without quotes. A handle alone is noise; a handle + quote is signal.
• Don't lead with your product. The first outreach message refers to their pain, not your feature.
• Don't mine the founder's followers. Fans ≠ pained users.

Related

• Playbook: [00-competitive-seed-pipeline](../../playbooks/00-competitive-seed-pipeline.md)
• Follow-up skill: [seed-user-pool-stratification](../seed-user-pool-stratification) — scores and segments the pool
• Follow-up skill: [personalized-cold-outreach](../personalized-cold-outreach) — writes the actual messages

Source & license

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

• Author: Celina-create
• Source: Celina-create/X-Studio
• License: MIT

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