How are the tasks categorized on the website

[lidh15]

In the tasks, repo, the 147 tasks are in these categories without overlapping:

  ┌──────────────────┬───────┐
  │     Category     │ Count │
  ├──────────────────┼───────┤
  │ log_analysis     │ 30    │
  ├──────────────────┼───────┤
  │ meeting_analysis │ 28    │
  ├──────────────────┼───────┤
  │ csv_analysis     │ 26    │
  ├──────────────────┼───────┤
  │ analysis         │ 12    │
  ├──────────────────┼───────┤
  │ research         │ 12    │
  ├──────────────────┼───────┤
  │ coding           │ 14    │
  ├──────────────────┼───────┤
  │ skills           │ 6     │
  ├──────────────────┼───────┤
  │ writing          │ 6     │
  ├──────────────────┼───────┤
  │ productivity     │ 8     │
  ├──────────────────┼───────┤
  │ integrations     │ 3     │
  ├──────────────────┼───────┤
  │ memory           │ 2     │
  ├──────────────────┼───────┤
  │ TOTAL            │ 147   │
  └──────────────────┴───────┘

while on the website, the 147 tasks are in the 8 categories, and they can be overlapped:

  ┌────────────────────┬──────────────────────┐
  │         ID         │        Label         │
  ├────────────────────┼──────────────────────┤
  │ code_devops        │ Code & DevOps        │
  ├────────────────────┼──────────────────────┤
  │ data_analysis      │ Data & Analysis      │
  ├────────────────────┼──────────────────────┤
  │ writing_content    │ Writing & Content    │
  ├────────────────────┼──────────────────────┤
  │ productivity       │ Productivity         │
  ├────────────────────┼──────────────────────┤
  │ research_knowledge │ Research & Knowledge │
  ├────────────────────┼──────────────────────┤
  │ security_ops       │ Security & Ops       │
  ├────────────────────┼──────────────────────┤
  │ core_agent         │ Core Agent           │
  ├────────────────────┼──────────────────────┤
  │ creative           │ Creative             │
  └────────────────────┴──────────────────────┘

what are the details about the 8 categories shown on the website?

[ScuttleBot]

👋 Hi @lidh15! I'm @olearycrew's OpenClaw bot, helping triage this repo.

Great question! The 8 website categories are a display-level grouping that maps from the frontmatter category field in each task file. Here's how it works:

Mapping Logic

The leaderboard uses a multi-layered approach (defined in lib/task-metadata.ts):

1. Explicit fallback table — Each task has a hardcoded website category assignment
2. Legacy category map — Old frontmatter categories get mapped:
   • api → data_analysis
   • validation → core_agent
   • calendar → productivity
   • research → research_knowledge
   • writing → writing_content
   • coding → code_devops
   • comprehension / context / complex → core_agent
3. Pattern matching — Task IDs are matched against regex patterns to infer category

The 8 Website Categories (with what maps to them)

Website Category	Frontmatter categories that map here
Code & DevOps	coding + tasks matching test_generation, k8s, cicd, dockerfile, git_rescue, etc.
Data & Analysis	csv_analysis, log_analysis + tasks matching csv, spreadsheet, financial, earnings, stock
Writing & Content	writing + tasks matching blog, humanizer, readme, commit_message, summary
Productivity	productivity + tasks matching calendar, email, todo, daily_summary
Research & Knowledge	research + tasks matching market_research, executive, polymarket, contract
Security & Ops	Tasks matching access_log, cve, security, gh_issue_triage
Core Agent	memory, skills + tasks matching sanity, memory, files, workflow, openclaw
Creative	Tasks matching image_gen, image_identification

Why They Can Overlap

The frontmatter categories are 1:1 (each task has exactly one), but the website allows tasks to potentially appear in multiple display categories through the pattern matching — though in practice most tasks land in exactly one website category.

The Key Difference

• Frontmatter categories (11 total in your table): Fine-grained, used for benchmark execution and grouping in the task repo
• Website categories (8 total): Coarser, user-facing groupings designed for the leaderboard UI filtering experience

The meeting_analysis frontmatter category, for example, maps to data_analysis on the website. The log_analysis tasks also map to data_analysis or security_ops depending on content.

Hope that clarifies! Feel free to ask if you want more detail on any specific mapping.