New competitor-intel skill: daily competitive intelligence from social media and e-commerce

[webup]

Migrated from issue #109, originally opened by @AIwork4me. Moved to a discussion for broader design conversation.

Summary

Add a new bundled skill, competitor-intel, that runs on a daily schedule to automatically fetch competitor information from major social media and e-commerce sites, producing a continuous, auditable competitive-intelligence digest.

Motivation

ClawMaster aims to make OpenClaw a genuinely useful personal assistant. The skill ecosystem already covers cost summaries (clawprobe-cost-digest) and content drafting (content-draft), but it lacks automated intelligence gathering aimed at business decisions.

For non-technical users in growth, product, and marketing, manually tracking competitors is time-consuming and easy to miss. A scheduled competitive-intel skill can:

• Automatically gather and summarize competitor public information daily, reducing information asymmetry.
• Aggregate data scattered across platforms into a structured digest, reducing manual curation.
• Run automatically via OpenClaw's scheduled-task mechanism — no manual trigger needed.
• Emit Markdown digests usable directly in team weekly reports or decision memos.

Proposed skill: competitor-intel

Directory layout

bundled-skills/competitor-intel/
├── SKILL.md                  # skill description and workflow guidance
├── _meta.json                # metadata (name, description, dependency declarations)
├── references/
│   └── platforms.md          # per-platform data-source notes and collection rules
└── scripts/
    ├── fetch-competitor-data.mjs   # core collection script
    ├── generate-digest.mjs         # digest generation
    └── schedule-config.mjs         # scheduled-task config helper

Data-source coverage

Social media

Platform	What we collect	How
Weibo	Competitor official-account posts, trending topics, engagement	Public API / page parsing
Xiaohongshu (XHS)	Competitor brand notes, seeded content, comment trends	Public page parsing
Douyin	Competitor short videos, livestream teasers, engagement	Public data endpoints
Twitter / X	Competitor official tweets, product announcements, community feedback	Public API
WeChat Official Accounts	Competitor articles, campaign announcements	RSS / public article links

E-commerce

Platform	What we collect	How
Taobao / Tmall	Competitor pricing, sales volume, review summaries	Public product search
JD	Competitor pricing, promotions, reviews	Public product search
Pinduoduo	Competitor pricing, sales ranking	Public product search
Amazon	Competitor pricing, review trends, BSR	Public product pages

Core features

1. Scheduled automated collection

   • Runs on OpenClaw's scheduled-task mechanism (cron), default daily at 08:00.
   • Supports --period day|week|month to pick a collection window.
   • Users can customize frequency and execution time.

2. Competitor configuration

   • Users declare tracked competitors in ~/.openclaw/competitor-intel.json.
   • Entries include: competitor name, keywords, linked accounts, target platforms, product URLs, etc.
   • Supports multiple competitors × platforms × keywords.

3. Structured collection

   • Each data source adapts to a unified AdapterResult<T> shape (consistent with the project's existing adapter pattern).
   • Collected data is cached locally at ~/.openclaw/cache/competitor-intel/ to avoid duplicate fetches.
   • Collection respects per-source rate limits.

4. Intelligent digest generation

   • Structured summaries organized by competitor × platform.
   • Categories: price changes, new products / content, marketing campaigns, sentiment trends.
   • Supports period-over-period comparison (consistent with clawprobe-cost-digest).
   • Markdown output directly usable for team weeklies.

Example config

{
  "competitors": [
    {
      "name": "Competitor A",
      "keywords": ["brand-name-A", "product-line-A"],
      "accounts": {
        "weibo": "@CompetitorAOfficial",
        "xhs": "CompetitorA",
        "twitter": "@CompetitorA"
      },
      "products": [
        { "platform": "taobao", "url": "https://..." },
        { "platform": "jd", "keyword": "Competitor A flagship product" }
      ]
    }
  ],
  "schedule": "0 8 * * *",
  "platforms": ["weibo", "xhs", "taobao", "jd", "twitter"]
}

SKILL.md sketch

---
name: competitor-intel
description: Daily automated collection of public competitor information from major social media and e-commerce platforms, producing a structured competitive-intelligence digest. Use when the user needs competitive intelligence tracking, competitor monitoring, or daily market digest.
metadata:
  openclaw:
    requires:
      anyBins:
        - node
---

# competitor-intel

Use this skill when the user needs automated competitive intelligence from
social media and e-commerce platforms — not manual searching.

This skill reads the user's competitor configuration, fetches public data
from configured platforms, and generates a structured digest with:
- Price changes and promotions
- New product / content releases
- Marketing campaign tracking
- Sentiment trend changes
- Period-over-period comparison

## Critical rule
This skill is guidance plus runnable scripts, not a callable tool name.
Do not call `competitor-intel` as if it were a built-in tool.
First use `read` to load this `SKILL.md`.
Then use `exec` to run the bundled Node scripts with `node`.

## Commands
# Collect competitor data
node ${SKILL_DIR}/scripts/fetch-competitor-data.mjs --config ~/.openclaw/competitor-intel.json

# Generate the daily digest
node ${SKILL_DIR}/scripts/generate-digest.mjs --period day --summary

# Configure the schedule
node ${SKILL_DIR}/scripts/schedule-config.mjs --cron "0 8 * * *"

Technical constraints (following project conventions)

• Node.js v20+ only — no Python, Bun, or other runtimes (per the project's dependency policy).
• No native binary dependencies — all collection logic is pure JS, relying on fetch / npm packages.
• Follow the adapter pattern — each platform adapter returns AdapterResult<T>, matching shared/adapters/.
• Collection scripts run via exec — same pattern as clawprobe-cost-digest; a skill is guidance + executable scripts.
• All user-visible text goes through i18n — tags in digest templates must be registered in zh.ts / en.ts / ja.ts.

Rollout plan

1. Phase 1 — foundation: scaffold the skill directory, config parsing, schedule wiring; implement 1–2 platforms (Weibo + Taobao) as a collection MVP.
2. Phase 2 — platform expansion: add more social media (XHS, Douyin, Twitter) and e-commerce platforms (JD, Pinduoduo, Amazon).
3. Phase 3 — intelligence enhancement: plug in an LLM for smarter summaries, sentiment analysis, and trend prediction; add webhook / notification push.

Acceptance criteria

• bundled-skills/competitor-intel/ directory complete, including SKILL.md, _meta.json, scripts/.
• At least two social-media + two e-commerce platforms supported.
• Scheduled automated execution (cron), default daily.
• Local caching of collected results with cross-period comparison.
• Structured Markdown digest: price changes, new products, campaigns, etc.
• Competitor config supports many competitors × platforms × keywords.
• All scripts depend only on Node.js v20+, no native binaries.
• User-visible text is i18n-ed (zh / en / ja).
• Includes __tests__/ coverage.
• npm test passes locally.

Related

• LLM Wiki — a persistent, compounding knowledge base module #49 (LLM Wiki — persistent knowledge base module) — collection results can persist into the knowledge module.
• clawprobe-cost-digest skill — reuses its digest-generation and period-over-period comparison patterns.