patent-prosecution-agent
Use this skill for responding to patent office actions, examiner rejections, and patent prosecution.
Skill by michaelboeding
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/michaelboeding-patent-prosecution-agent
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: legal, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/patent-prosecution-agent
What's inside
- Skill name
patent-prosecution-agent- Description
- Use this skill for responding to patent office actions, examiner rejections, and patent prosecution. Triggers: "office action", "examiner rejection", "patent rejection", "102 rejection", "103 rejection", "101 rejection", "112 rejection", "respond to examiner", "amend claims", "narrow claims", "office action response", "patent prosecution", "final rejection", "non-final rejection", "restriction requirement", "appeal brief", "PTAB appeal", "RCE", "request for continued examination", "after-final amendment", "examiner interview", "claim amendment", "overcome rejection", "distinguish prior art", "traverse rejection", "patent response" Outputs: Office action analysis, response strategy, amended claims, arguments/remarks, full office action response, appeal brief. DISCLAIMER: This is informational only, not legal advice. Consult a licensed patent attorney.
- License
- MIT
- Instructions
- 387 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Files (7)
- SKILL.md
- argument-writer.md · agents/
- claim-amendment-drafter.md · agents/
- office-action-analyzer.md · agents/
- prior-art-distinguisher.md · agents/
- prosecution-strategist.md · agents/
- markdown_to_pdf.py · scripts/
SKILL.md (first lines)
# Patent Prosecution Agent Analyze patent office actions and draft responses to examiner rejections. **This is a companion skill to `patent-lawyer-agent`.** Use `patent-lawyer-agent` for initial patent drafting and strategy. Use this skill after the USPTO (or other patent office) issues an office action rejecting or objecting to your claims. **This skill uses 5 specialized agents** that analyze office actions from different perspectives and draft complete responses. ## What It Produces | Output | Description | |--------|-------------| | **Office Action Analysis** | Breakdown of each rejection type, cited references, and affected claims | | **Response Strategy** | Recommended approach for each rejection (amend, argue, or both) | | **Amended Claims** | Rewritten claims with proper markup showing additions and deletions | | **Arguments/Remarks** | Legal arguments distinguishing the invention over cited prior art | | **Full Response** | Complete office action response document ready for attorney review | | **Appeal Brief** | PTAB appeal brief for final rejections | ## Prerequisites - Web access for searching cited prior art references - `pip install markdown weasyprint` - For PDF generation ## How It Works **This agent is request-driven.** Tell it what you need and it uses whichever agents are required: | You Ask | What Happens | |---------|--------------| | "I got an office action, help me respond" | Full analysis + strategy + response drafting | | "Analyze this office action" | Parses rejections, identifies issues, maps claims | | "How should I respond to this 103 rejection?" | Strategy recommendation for the specific rejection | | "Amend my claims to overcome this rejection" | Drafts narrowed/amended claims with markup | | "Write arguments to distinguish over the cited art" | Drafts remarks section with legal arguments | | "I got a restriction requirement" | Analyzes and recommends claim election | | "I got a final rejection, what are my options?" | RCE vs appeal vs after-final amendment analysis | | "Prepare an appeal brief" | Drafts PTAB appeal brief | **You don't need to follow a rigid workflow.** The agent uses whichever specialized agents are needed for your request. --- ## Request Types ### 1. Full Office Action Response "I received an office action, help me respond" / "Draft a response to this rejection" **Use the `AskUserQuestion` tool for each question below.** Do not just print questions in your response — use the tool to create interactive prompts. **Q1: Office Action** > "I'll help you respond to this office action! > > **First — share the office action details.** > > You can: > - Paste the full office action text > - Provide a summary of the rejections > - Upload the office action document > > *(Include the rejection types, cited references, and affected claims)*" *Wait for response.* **Q2: Your Claims** > "Now share your **current claims** as filed (or as last amended). > > *(Paste the full claim set so I can analyze what the examiner is rejecting)*" *Wait for response.* **Q3: Invention Context** > "Any additional context about your invention? > > - What makes your invention different from the cited art? > - Are there features the examiner may have missed? > - Any preferred scope you want to maintain? > - Or say 'none' and I'll work from the claims alone" *Wait for response.* **What it does:** 1. Parses and categorizes every rejection in the office action 2. Maps each rejection to affected claims and cited references 3. Develops a response strategy for each rejection 4. Drafts amended claims (if amendments are recommended) 5. Writes arguments/remarks distinguishing over prior art 6. Assembles complete response document **Uses agents:** Office Action Analyzer, Prosecution Strategist, Claim Amendment Drafter, Argument Writer, Prior Art Distinguisher **Output:** Complete office action response in Markdown + PDF --- ### 2. Office Action Analysis Only "Analyze this office action" / "What is the examiner saying?" **What it does:** 1. Parses the office action into structured components 2. Categorizes each rejection by type (101, 102, 103, 112) 3. Identifies cited prior art references 4. Maps rejections to specific claims 5. Highlights the examiner's key reasoning 6. Identifies the strongest and weakest rejections **Uses agents:** Office Action Analyzer --- ### 3. Response Strategy "How should I respond to this rejection?" / "What's the best strategy here?" **What it does:** 1. Analyzes the rejection(s) 2. Evaluates strength of examiner's position
From michaelboeding/skills, MIT.
More from michaelboeding/skills
- 0.656score
expired-patent-scout
michaelboeding
Find recently expired patents relevant to specified brands, products, or technology areas and develop evidence-backed product opportunity leads.
SkillAFreshness: A★ 30Claude Code +9 - 0.656score
mobile-parity-check
michaelboeding
Compare the UI and UX of iOS and Android apps in either direction.
SkillAFreshness: A★ 30Claude Code +9 - 0.656score
search-trend-product-scout
michaelboeding
Discover and rank product ideas in a category using search trends, customer problems, and competing products.
SkillAFreshness: A★ 30Claude Code +9 - 0.616score
chart-generation
michaelboeding
Use this skill for generating data-driven charts and visualizations using Python.
SkillBFreshness: B★ 30Claude Code +9 - 0.501score
image-generation
michaelboeding
Use this skill for any image-related AI generation or editing task.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
auto-permissions-review-install
michaelboeding
Install an AI-powered permission review hook that uses Claude Haiku to evaluate tool call safety before execution.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
patent-lawyer-agent
michaelboeding
Use this skill for intellectual property guidance, prior art research, patent strategy, and drafting patent applications.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
cmo-agent
michaelboeding
AI Chief Marketing Officer that analyzes your website and deploys autonomous marketing agents across SEO, content, Reddit, Hacker News, X/Twitter, and AI…
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
icon-generation
michaelboeding
Generate app icons with transparent backgrounds.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
brand-research-agent
michaelboeding
Use this skill to analyze a brand from their website before creating content.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
music-generation
michaelboeding
Use this skill for AI music generation.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
slide-generation
michaelboeding
Use this skill to create presentation slides from structured content.
SkillDFreshness: D★ 30Claude Code +9
Signals
- Stars
- 30
- Forks
- 2
- Open issues
- 0
- Last commit (this folder)
- Apr 12, 2026 (6 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/michaelboeding/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from michaelboeding
- 0.656score
mobile-parity-check
michaelboeding
Compare the UI and UX of iOS and Android apps in either direction.
SkillAFreshness: A★ 30Claude Code +9 - 0.656score
expired-patent-scout
michaelboeding
Find recently expired patents relevant to specified brands, products, or technology areas and develop evidence-backed product opportunity leads.
SkillAFreshness: A★ 30Claude Code +9 - 0.656score
search-trend-product-scout
michaelboeding
Discover and rank product ideas in a category using search trends, customer problems, and competing products.
SkillAFreshness: A★ 30Claude Code +9 - 0.616score
chart-generation
michaelboeding
Use this skill for generating data-driven charts and visualizations using Python.
SkillBFreshness: B★ 30Claude Code +9 - 0.501score
auto-permissions-review-install
michaelboeding
Install an AI-powered permission review hook that uses Claude Haiku to evaluate tool call safety before execution.
SkillDFreshness: D★ 30Claude Code +9 - 0.501score
brand-research-agent
michaelboeding
Use this skill to analyze a brand from their website before creating content.
SkillDFreshness: D★ 30Claude Code +9
Cite this page
patent-prosecution-agent skill. Agent Mart. https://agentmart.dev/skills/michaelboeding-patent-prosecution-agent (last updated 2026-10-24)