← 返回 Skills 市场
agents365-ai

Asta Skill

作者 Agents365.ai · GitHub ↗ · v0.2.7 · MIT-0
cross-platform ⚠ suspicious
172
总下载
1
收藏
0
当前安装
5
版本数
在 OpenClaw 中安装
/install asta-skill
功能描述
Domain expertise for Ai2 Asta MCP tools (Semantic Scholar corpus). Intent-to-tool routing, safe defaults, workflow patterns, and pitfall warnings for academi...
使用说明 (SKILL.md)

Asta MCP — Academic Paper Search

Asta is Ai2's Scientific Corpus Tool, exposing the Semantic Scholar academic graph over MCP (streamable HTTP transport). This skill tells agents which Asta tool to call for which intent, and how to compose them into useful workflows.

Update Notifications

If \x3Cskill_directory>/.last_version_check is missing or >24h old, fetch the version from https://raw.githubusercontent.com/Agents365-ai/asta-skill/main/SKILL.md, compare to the local version, then write the current Unix timestamp to .last_version_check. If remote is newer, tell the user once: 📢 asta-skill X.Y.Z → A.B.C. Run 'git -C \x3Cskill_directory> pull' to update. Silently skip on any failure. Never run git pull yourself.

Prerequisite Check

Before invoking any tool, verify the Asta MCP server is registered in the host agent. Tool names will be prefixed by the MCP server name chosen at install time (commonly asta__\x3Ctool> or mcp__asta__\x3Ctool>). If no Asta tools are visible, direct the user to the Installation section below.

Tool Map — Intent → Asta Tool

User intent Asta tool Notes
Broad topic search search_papers_by_relevance Supports venue + date filters
Known paper title search_paper_by_title Optional venue restriction
Known DOI / arXiv / PMID / CorpusId / MAG / ACL / SHA / URL get_paper Single-paper lookup
Multiple known IDs at once get_paper_batch Batch lookup — prefer over N sequential get_paper calls
Who cited paper X get_citations Citation traversal with filters, paginated
Find author by name search_authors_by_name Returns profile info
An author's publications get_author_papers Pass author id from previous call
Find passages mentioning X snippet_search ~500-word excerpts from paper bodies

Search/citation tools accept publication_date_range (format YYYY-MM-DD:YYYY-MM-DD; year shorthand like "2021:", ":2015-01", "2015:2020" is also accepted) and venues (comma-separated) filters, plus fields for field selection — pass them whenever the user's intent constrains scope (e.g., "recent", "since 2022", "at NeurIPS").

⚠️ fields parameter — avoid context blowups

get_paper / get_paper_batch accept a fields string. Never request citations or references via fields — a single highly-cited paper (e.g. Attention Is All You Need) returns 200k+ characters and will overflow the agent's context window. Use the dedicated get_citations tool for forward citations (it paginates). Asta does not provide a dedicated get_references tool — to retrieve a paper's reference list, use get_paper with fields=references only for papers you know have a small reference list (typically \x3C 100).

Safe default fields for get_paper:

title,year,authors,venue,tldr,url,abstract

Add journal, publicationDate, fieldsOfStudy, isOpenAccess only when needed.

Workflow Patterns

Pattern 1 — Topic Discovery

  1. search_papers_by_relevance(keyword, publication_date_range="\x3Ccurrent_year-5>:", venues=?) → initial hits (compute the lower bound from today's date — e.g., in 2026 pass publication_date_range="2021:"; adjust or drop the filter if the user asks for older work)
  2. Rank/present top N by citationCount + recency
  3. Offer follow-ups: get_citations on the most influential, or snippet_search for specific claims

Pattern 2 — Seed-Paper Expansion

  1. get_paper(DOI|arXiv|...) → verify seed
  2. get_citations(paperId) → forward expansion
  3. Optionally search_papers_by_relevance with seed title terms for sideways discovery
  4. Deduplicate by paperId before presenting

Pattern 3 — Author Deep-Dive

  1. search_authors_by_name(name) → pick correct profile (disambiguate by affiliation)
  2. get_author_papers(authorId) → full publication list
  3. Filter client-side by topic keywords or date

Pattern 4 — Evidence Retrieval

  1. snippet_search(claim_query) → find passages making/supporting a claim
  2. For each hit, optionally get_paper(id) for full metadata

Output & Interaction Rules

  • Always report total count and which tool was used.
  • Present top 10 as a table (title, year, venue, citations), then details for the most relevant.
  • If the user writes in Chinese, present summaries in Chinese; keep titles in original language.
  • After results, offer: Details / Refine / Citations / Snippet / Export / Done.

Critical Rules

  • Prefer batched intent over ping-pong. If the user's question needs two independent lookups, issue them as parallel MCP tool calls in one turn, not sequentially.
  • Never guess IDs. If a user gives a fuzzy title, use search_paper_by_title before get_paper.
  • Respect rate limits. An API key buys higher limits but not unlimited — stop expanding citation graphs beyond what the user asked for.
  • Do not fabricate fields. If Asta returns null abstract or venue, say so rather than inventing.

Handling Asta responses

Situation What to do
Empty abstract Not all corpus papers have full text — use snippet_search, or fall back to title + TLDR
Author disambiguation uncertain Inspect affiliations in search_authors_by_name results before calling get_author_papers
429 Too Many Requests Back off; batch with get_paper_batch instead of sequential get_paper calls
安全使用建议
This skill is coherent and appears to do what it says: it helps agents use Ai2's Asta MCP tools and only needs your ASTA_API_KEY. Before installing: (1) ensure you trust the ASTA API key destination and that you store ASTA_API_KEY in a secure place (environment variables are typical); (2) be aware the skill can be implicitly invoked by the host when you ask about papers/citations; (3) if you want to avoid any outbound network fetches beyond the Asta MCP server, note the SKILL.md documents an optional update-check that fetches a SKILL.md from raw.githubusercontent.com and writes a .last_version_check file — you can disable or ignore that behavior or inspect/implement it on the host side; (4) if you need stricter isolation, confirm your host enforces MCP server registration and does not allow arbitrary outbound calls by skills. Overall: safe to install if you trust the Asta/Allen AI endpoint and the ASTA_API_KEY request.
能力标签
cryptocan-make-purchasesrequires-sensitive-credentials
能力评估
Purpose & Capability
Name, description, and declared requirements align: the skill is an instruction pack for Ai2's Asta MCP tools and only requires ASTA_API_KEY (used as the x-api-key header to the Asta MCP server). No unrelated binaries, credentials, or platform access are requested.
Instruction Scope
SKILL.md stays within the scope of guiding MCP usage, intent→tool routing, safe defaults, and workflows. One inconsistency: the header claims "The skill does not make HTTP calls itself," yet the Update Notifications section instructs the agent to fetch https://raw.githubusercontent.com/.../SKILL.md to check versions and to write a .last_version_check file in the skill directory. The update-check is a limited external fetch for version metadata and writing a local timestamp; this is not required for Asta functionality but is a documented optional behavior.
Install Mechanism
Instruction-only skill (no install spec, no code files to execute). Installation guidance is limited to cloning the repo into normal skill directories. No remote installers, archived downloads, or executables are installed by the skill itself.
Credentials
Only one environment variable is required: ASTA_API_KEY, which directly corresponds to the Asta MCP x-api-key header — proportional and expected for this integration. No additional secrets or unrelated credentials are requested.
Persistence & Privilege
The skill is not force-included (always:false). It requests only permission to write a small per-skill file (.last_version_check) in its own directory for optional update-notifications; it does not request system-wide config changes or other skills' credentials. The agents/openai.yaml allows implicit invocation (allow_implicit_invocation: true), which means the skill can be triggered automatically by the host when relevant — this is typical for skills but users should be aware of it.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install asta-skill
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /asta-skill 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.2.7
fix: use publication_date_range (was year=); condense update-notification block
v0.2.4
Add opencode and pi-mono platforms; rename OpenClaw → OpenClaw/ClawHub; rolling 5-year window default in Pattern 1.
v0.2.2
Add daily auto-update, contributing section, remove GitHub Actions
v0.2.1
asta-skill v0.2.1 - Added LM Studio installation section to compatibility and setup instructions. - Updated metadata to version 0.2.1. - Documentation improvements in README files and SKILL.md for broader tool compatibility and usage clarity.
v0.2.0
asta-skill 0.2.0 - Added comprehensive SKILL.md documentation covering usage, workflow patterns, tool mapping, and installation instructions. - Clarified compatibility requirements and MCP server configuration for multiple platforms and hosts (Claude Code, Codex, Cursor, Windsurf, Hermes, OpenClaw). - Provided best-practice guidelines for safe field selection and batching queries to avoid context overflow and rate limits. - Outlined troubleshooting steps and differences compared to semanticscholar-skill. - Specified metadata and environment variable requirements for OpenClaw and other agents.
元数据
Slug asta-skill
版本 0.2.7
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 5
常见问题

Asta Skill 是什么?

Domain expertise for Ai2 Asta MCP tools (Semantic Scholar corpus). Intent-to-tool routing, safe defaults, workflow patterns, and pitfall warnings for academi... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 172 次。

如何安装 Asta Skill?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install asta-skill」即可一键安装,无需额外配置。

Asta Skill 是免费的吗?

是的,Asta Skill 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Asta Skill 支持哪些平台?

Asta Skill 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Asta Skill?

由 Agents365.ai(@agents365-ai)开发并维护,当前版本 v0.2.7。

💬 留言讨论