wentorai/research-claw

Output Cards

>- JSON schema reference for 6 structured output card types used by the Research-Claw dashboard: paper_card, task_card, progress_card, approval_card, file_card, monitor_digest. Read this before outputting any structured card for the first time in a session.

Installation

$ npx skills add wentorai/research-claw --skill output-cards

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from wentorai/research-claw · top by installs.

npx skills add wentorai/research-claw

Browse all from wentorai/research-claw

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 851
License LICENSE
Default branch main
Open issues 0
Status Active

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,975 B
  • docs SUMMARY.md 274 B

History

  1. First recorded snapshot · 10 installs

SKILL.md

<!-- SKILL MAINTENANCE NOTES: - Dashboard 用 JSON.parse() 解析卡片内容 - 字段变更必须同步 dashboard/src/components/chat/ 中的渲染组件 - workspace file facts and supported raw search results are projected automatically - papercard is an Agent-deliberate highlight, not a raw-result dump - approvalcard 必须包含 approval_id(来自 exec.approval.requested) -->

Output Cards

Use fenced code blocks with the card type as the language tag only for Agent-authored card semantics. Content MUST be valid JSON — the dashboard parser uses JSON.parse(). Research-Claw automatically presents successful Workspace file facts and supported raw literature results; do not copy tool JSON merely to make those cards appear. Keep ordinary paths and safe links in the surrounding prose for non-Dashboard channels.

paper_card

ONLY for real academic publications deliberately highlighted by the Agent — from API queries, library_search, or user-identified papers. NEVER for concepts, tools, non-scholarly content, or every raw search hit. A raw result is only retrieved; the fence does not prove it was read, cited, saved, or verified.

Required: type, title, authors (string[]). Optional: venue, year, doi, url, arxivid, abstractpreview, readstatus ("unread"|"reading"|"read"|"reviewed"), libraryid, tags.

task_card

Required: type, title, tasktype ("human"|"agent"|"mixed"), status ("todo"|"inprogress"|"blocked"|"done"|"cancelled"), priority ("urgent"|"high"|"medium"|"low"). Optional: id, description, deadline (ISO 8601), relatedpapertitle, relatedfilepath.

progress_card

Required: type, period, papersread, papersadded, taskscompleted, taskscreated. Optional: writingwords, readingminutes, highlights (max 5).

approval_card

Required: type, action (string), context (string), risklevel ("low"|"medium"|"high"). Required (for exec approvals): approvalid from exec.approval.requested — without it, dashboard buttons are non-functional. Optional: details (must be a JSON object, not a string — e.g. {"paper_count": 7}).

file_card

Successful workspacesave, workspaceexport, workspaceappend, and workspacedownload results are presented automatically. State the ordinary workspace-relative path in prose; no fence is required. Legacy fences remain compatible, but NEVER fabricate a path.

monitor_digest

Required: type, monitorname, sourcetype (free-form), target, total_found, findings (array of {title, url?, summary?}, max 10). Optional: schedule.