add social-signal-scraping skill

[urugondavishnu]

social-signal-scraping

Scrape social signals (Reddit, Twitter/X, Product Hunt) about any company, product, or topic using TinyFish. Returns structured JSON with sentiment analysis, pain points, positive feedback, and trending topics.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 2f6ac7a6-9657-4edb-a14d-ac1e532efa06

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[simantak-dabhade]

Hey @urugondavishnu! Great concept, two things to clean up:

1. Remove the keyword-based sentiment classification
The skill includes a manual keyword scoring system (positive/negative word lists, ratio scoring, etc.) — but this is a skill running inside Claude, which is an LLM. Claude can assess sentiment directly from the content without any of this. Hardcoding word lists and scoring formulas is overengineering the wrong layer. Just instruct Claude to classify sentiment as part of the synthesis step and it'll do a far better job than keyword matching ever could.

2. Remove the TypeScript implementation reference
The bottom half of the SKILL.md contains a full TypeScript code scaffold — runTinyFish(), scrapeReddit(), aggregateResults(), etc. A skill isn't a codebase — it's instructions for an AI agent. Claude doesn't need application code to execute the skill, it needs clear natural language instructions on what to do and in what order. This TypeScript block belongs in a separate app project, not in a SKILL.md.

Think of the skill purely as: what should Claude do, step by step, using the TinyFish CLI. Everything else is noise that inflates the skill size without helping the agent.

[urugondavishnu]

Thanks for the feedback @simantak-dabhade ! I’ve made the requested changes:

• Removed the keyword-based sentiment classification and now rely on Claude to infer sentiment directly from the content.
• Removed the TypeScript scaffold and converted the skill into clear step-by-step instructions using the TinyFish CLI.

The skill is now purely instruction-based and focuses only on what Claude should do at each step. LMK if you'd like any further tweaks.

[simantak-dabhade]

Hey! Thanks for the submission — the SKILL.md is well-structured and the concept is solid. A few things need to be addressed before we can merge:

Critical:

1. Nitter is dead — The skill routes Twitter/X scraping through nitter.net, but Nitter shut down in mid-2023 and most mirrors are unreliable. This leg of the skill won't work. Please either:

   • Use x.com/search directly (accept that it may need login/stealth)
   • Or remove Twitter as a platform and document the limitation

2. Product Hunt via Google site: search is fragile — Searching Google for site:producthunt.com then clicking into results is a multi-hop chain that'll hit CAPTCHAs and rate limits. Use Product Hunt's own search directly: producthunt.com/search?q=<query>

Minor:

3. Add a .env.example with TINYFISH_API_KEY= — repo convention
4. The --async flag usage references tinyfish agent run get <run_id> for polling but doesn't show the actual polling syntax. Please clarify the polling loop or command.
5. Sentiment is classified twice (once during extraction in Step 1, again in Step 3). Pick one and remove the redundancy.

Once the Nitter and Product Hunt issues are fixed, this is close to ready!