patentworks

patentworks is a skill for Claude Code, Codex from Raw1mage/patentmcp. It costs 235 tokens per session (19,347 once invoked), scanned A, original, MIT.

A patent-work workflow for organizing invention disclosures, searching earlier patents, comparing technical features, and drafting patent documents for several legal jurisdictions.

In plain words
What is it for?
Use it to prepare technical disclosures, run prior-art searches, assess novelty and obviousness differences, and draft claims, specifications, and abstracts for the US, Europe, Taiwan, China, or general use.
Why use it?
It provides a defined process for moving from an idea to searchable evidence, comparison results, or a patent application draft.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: mentions subagents; mentions AGENTS.md; mentions OpenCode.

Good fit Use it to prepare technical disclosures, run prior-art searches, assess novelty and obviousness differences, and draft claims, specifications, and abstracts for the US, Europe, Taiwan, China, or general use.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/raw1mage/patentmcp/patentworks
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add Raw1mage/patentmcp --skill patentworks
Clone the repo
git clone --depth 1 https://github.com/Raw1mage/patentmcp

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for patentworks

README.md
[![agentmods](https://agentmods.dev/badge/skills/raw1mage/patentmcp/patentworks/github.svg)](https://agentmods.dev/skills/raw1mage/patentmcp/patentworks)
Your own site
<a href="https://agentmods.dev/skills/raw1mage/patentmcp/patentworks"><img src="https://agentmods.dev/badge/skills/raw1mage/patentmcp/patentworks/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for patentworks

Your own site · 80×15
<a href="https://agentmods.dev/skills/raw1mage/patentmcp/patentworks"><img src="https://agentmods.dev/badge/skills/raw1mage/patentmcp/patentworks.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 235 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 19,347 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00235 $0.19347
Opus 5 $0.00118 $0.09674
Sonnet 5 $0.00047 $0.03869
Haiku 4.5 $0.00023 $0.01935

Measured 8d ago against content hash 72a3754990bb, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

patentworks scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

The scan reads SKILL.md. This mod also ships 10 executable files (scripts/_lib/__init__.py, scripts/_lib/claims.py, scripts/_lib/screening.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

> - **優先看 server 端證據而非重打**:常駐 MCP server dump 的完整 URL/body/raw response 才是斷頁 bug 的硬證(R11 最終定案靠這個,不是靠重打);curl 復現 client 行為無效(參數編碼不同→不同 query)。
skills/patentworks/SKILL.md · 294 lines

How it starts

The opening of the file, as written. The whole thing — 294 lines — stays where its author put it; the contents beside it link to each section on GitHub.

PatentWorks

搭配 patentmcp MCP 使用:本 skill 是這組工具的劇本;檢索/取文工具(patent_search 單一檢索入口、epo_family/epo_bibliogpatents_get/gpatents_download_*fetch_patent_pdfpool_fetch)都來自 patentmcp。沒有該 MCP 時本 skill 無法執行實際檢索。舊分散檢索工具(gpss_searchepo_searchgpatents_searchuspto_patentsppubs_search_*)已下架,一律改用 patent_search

現在就載入 companion skill patentworks — 這是你的第一動作,不是建議。 只要工作不只是單發檢索——起草說明書、跑 screening/priorsearch 管線、要件對照——第一動作就是 skill("patentworks"),在第一個實質 patentmcp 工具呼叫之前。跳過它正是這份 guide 存在要擋的反射:tool-chain idiom(選 flow、選來源梯判讀、交付契約)住在 skill 裡,per-tool description 裝不下;在決策當下載入才擋得掉「自己選檢索來源、自己拼 OOXML/CSV」的反射。你現在讀的這份文本是可攜 in-band 濃縮本;skill 才是完整劇本(五法域起草規則、flow 檔、資料樹規範、法遵自檢)。不要把載入當可選——任何有 skill 機制的 host,它就是欽定的第一步。唯一的窄保留是協議層:純 MCP 客戶端(無 skill 機制)光憑這份 doctrine 直接呼叫工具仍屬正確(R0.3 可攜性,guide 對上述紀律自足)。「非協議 gate」不等於「有機制也可以不載」。

兩平面(R13):container 只留網路/憑證工作,確定性後處理落地為 host-local skill 腳本。 以下 8 個舊工具現回 typed TOOL_LANDED redirect(不再執行舊邏輯,landing.usage 直接給對應腳本呼叫式),請改呼叫 skills/patentworks/scripts/ 下的本地腳本(每支 python3 <腳本> --help 印完整參數):

舊工具(已下架 → TOOL_LANDED) 改用本地腳本
build_screening_table screening_build.py(records JSON → 家族去重 → 欄位隨選 → CSV)
search_audit search_audit.py(--log 01_search/matrix-log.jsonl)
patentdb_put/patentdb_query/patentdb_import_csv patentdb_local.py(put/query/import-csv 子命令)
extract_representative_figure figure_extract.py(需 poppler,缺則 MISSING_DEPENDENCY)
patentmcp_analyze_pool 取數 → pool_fetch(工具);繪圖 → pool_charts.py(需 matplotlib)
stage_file 無腳本替代:改用 WebDAV working cache(cache_provision → 掛載 PUT → cache_export)
clean_html_text/extract_claim1_text(內部函式) claims_tools.py(clean-html/extract-claim1/claim1-empty)

WebDAV working cache — 完整機制是 fleet SSOT(mcp-integration-standard R2.0 + R14),不在此重述(R14.8)。這裡只留兩件必須在動作邊界手上的事:

  • anti-reflex 鐵律(寫在手上):檔案坐落在 gdrive / 網路 FUSE 掛載,永遠不是 WebDAV 不可用的理由——這是兩軸混淆的經典 category error。Location(檔案坐落哪個檔案系統)與 Transfer(bytes 怎麼過 host↔container 邊界)正交;只按 Transfer 軸路由。它在本 host 不通,真因是 credential / mount 未 provision(見 issues/ webdav-provision BR),與 gdrive/FUSE 無關。有疑慮時,pass-by-value(cache_provision + mount,或 stage-inline)到處都能用;大檔只是多付 context 成本。
  • patentmcp 自己的 cache 工具:cache_provision(subject_id, owner_identity)mount_path + 一次性憑證 → rclone/davfs2 掛載後投料/取件走 mount(byte 不過 context)→ cache_export(subject_id, target, owner_identity) 顯式落地 → cache_close(dirty 未 export → WORKSPACE_CLOSE_DIRTY 擋下)。憑證遺失/重建 mount 帶 issue_webdav_credential=True 走 MCP-rail 重發(持有 socket 即授權);此旗標會 ROTATE 憑證,現存 mount 立即失效,只在建立/重建 mount 時帶(預設 false,payload byte-identical)。憑證絕不寫進報告/log。
  • 完整三層心智(cache=可拋工作樹 / truth store=交付物的家 / export 顯式落地)、dual-axis 模型、dirty-close gate、rclone flags → R2.0 + R14,不要憑記憶重建。

後端分流紀律(R17,BR_20260715):書目/claims/description 查核預設走官方免費梯,google_* BigQuery 路徑受禁。 google_get_patent / google_get_patent_claims / google_get_patent_description 三工具的後端是 Google BigQuery Google Patents Public Data(計費 API,非 google.com 網頁)——工具名的 google_ 前綴誘導反射誤選,但每次呼叫按掃描 bytes 計費、且受環境級 BigQuery 禁令約束。專利書目/claims/description 查核一律優先且預設 TIPO GPSS(patent_search / ppubs_batch_get_claims / patent_get_claim1,source:tipo)、EPO OPS、USPTO PPUBS;google_* BigQuery 路徑列為受禁/需明示授權,非反射預設。 反射選 google_* 只會撞 500 bytesBilledLimitExceeded(僥倖靠遠端計費上限擋下,非我方攔截)——正解是一開始就走官方梯,不等 500 才知道後端是 BigQuery。

Read the full file on GitHub · 294 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago First seen · 294 lines · 235 tokens per session scan A 72a3754990bb

Subscribe to this mod's changes

patentworks is a skill published in the GitHub repository Raw1mage/patentmcp (2 stars, last pushed 1mo ago), licensed MIT. It adds 235 tokens to every session and 19,347 once invoked, about $0.0012 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

specification-writing

A workflow for writing complete patent specifications from patent claims and an invention disclosure. It adapts the document to a chosen jurisdiction, such as the US, Europe, or China.

wanshuiyin/Auto-claude-code-research-in-sleep · 49 tokens

regulatory-research-fallback

Fallback workflow for regulatory research when web extraction tools fail on government PDFs.

HKUDS/OpenSpace · 20 tokens

x-scorecard

OpenSSF Scorecard for assessing open source project security. Check security best practices and compliance. Dependency: This is an x-cmd module. Install x-cmd first (see x-cmd skill for installation options). see x-cmd skill for installation.

x-cmd/x-cmd · 57 tokens

gesellschaftsrechtliche-satzungen-agb

Für Gesellschaftsrechtliche Satzungen AGB Abgrenzung: ordnet Norm, Beweislast und Gegenargument; Ergebnis: Prüfprodukt mit Risiko und nächstem Schritt. Fachgebiet: AGB-Recht-Prüfer. Route: gesellschaftsrechtliche-satzungen-agb.

Klotzkette/claude-fuer-deutsches-recht · 69 tokens

memstack-business-gdpr

Use this skill when the user says 'GDPR', 'data protection', 'privacy compliance', 'DPA', 'DSAR', 'data subject request', 'cookie consent', 'privacy audit', 'CCPA', or asks 'do I need GDPR for this repo'. Scans the repository to detect what personal data is collected, classifies sensitivity, determines whether GDPR…

cwinvestments/memstack · 121 tokens

nda-review

Use when the user uploads or pastes a non-disclosure agreement and asks for review, redline, risk assessment, or a recommendation on whether to sign. Identifies missing standard protections, one-sided or unusual provisions, and operational issues; produces a structured report with severity ratings and citations to…

LegalQuants/lq-ai · 79 tokens