echo-test
Test-only skill: echoes the input message back verbatim and writes it to an artifact file. Used by Skill Host unit tests.
- License 许可
- MIT
- 最近更新
- 2026-06-07
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- echo-test
- domain
- test
- version
- 1.0.0
- triggers
- [ "echo", "echo message", "test echo" ]
- description
- Test-only skill: echoes the input message back verbatim and writes it to an artifact file. Used by Skill Host unit tests.
02 资源文件 1
- 脚本scripts/echo.py707 B
03 SKILL.md 正文
Echo Test Skill
A minimal skill used only in the Skill Host test suite.
Tools
echo: return themessageargument verbatim and write it toartifacts/echo.txt.
04 同仓库更多 Skills
writing-core
Enforce de-AI-ification language rules and Markdown formatting standards for academic prose. Maintains banned-word lists (mechanical transitions, hollow emphasis phrases, hollow modifiers, subjective lead-ins), paragraph 3-section rule (topic / supporting / closing), list-to-prose conversion, and three-pass quality checks (structure / language / typography). Use when finalising any paper section, after paper-writing produces prose, before paper-revision iterates, or when the user says "去AI化", "polish prose", "remove AI flavour", "去除AI味", "正文太机械", or asks for style audit. Always paired with `scripts/style_check.py` for mechanical verification.
writing-chapters
Write paper chapters one at a time under a strict Chapter Agent Contract: type-routing (Intro/Related Work → evidence-driven-writing; Methodology → input-to-output flow; Results/Discussion → experiment-results-planning), anti-enumeration 5-pattern prose, min_chars enforcement from `plan/chapter-architecture.md`, two-stage review (spec compliance + quality), and structured handoff status (DONE / DONE_WITH_CONCERNS / NEEDS_CONTEXT / BLOCKED). Use when the user says "写第 N 章 / 写 introduction 章 / 写方法章 / 写结果章 / 一章一章地写 / chapter-by-chapter / write methodology / write results / 逐章节写作", or when paper-writing / paper-orchestration dispatches a single-chapter task.
verification
Universal "claim → evidence" gate. Forbids announcing completion without running a verification command and reading its full output. Maintains a Red-Flag list (apologetic / hedging / fatigue language) and an Excuses-vs-Facts table to mechanically intercept "should be done" / "looks fine" / "I'm pretty sure" style finalisations. Use when the agent is about to claim chapter complete / references verified / skill finished / search done / repo healthy / paper ready, or when the user says "确认 / 验证 / 真的写完了吗 / verify / are you sure". Distinct from `claim-verification` which audits paper-internal numerical claims; this skill audits skill-completion claims at every level.
survey-writing
Generate structured literature survey reports from reading notes and findings. Use when concluding a research investigation, producing a comparison table, summarising findings, or identifying research gaps.
survey-table
维护论文调研总表和生成方向级调研报告。调研单篇论文时按表格列定义进行结构化分析; 调研一个小方向时生成包含论文表格 + 分类总结的完整报告。所有优秀论文统一维护在 data/survey/paper-table.md 主表中。
skill-creator
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.