patent-novelty-check

name: "patent-novelty-check" description: "Assess patent novelty and non-obviousness against prior art. Use when user says "专利查新", "patent novelty", "可专利性评估", "patentability check", or wants to evaluate if an invention is patentable." allowed-tools: Bash(*), Read, Write, Edit, Grep, Glob, Agent, WebSearch, WebFetch

Override for Codex users who want Claude Code CLI, not a second Codex agent, to act as the reviewer/helper. Install this package after skills/skills-codex/*.

Whenever the upstream skill asks for an external reviewer/helper, write the complete focused prompt to $PROMPT_FILE. For a one-shot independent review, run:

claude -p --dangerously-skip-permissions --output-format json --model opus --effort max < "$PROMPT_FILE" | tee "$RAW_REVIEW_JSON"

For multi-round reviewer discussion, keep automation non-interactive but preserve continuity with --session-id on the first call and --resume on follow-up calls; see ../shared-references/claude-cli-review.md.

Patent Novelty and Non-Obviousness Check

Assess patentability of: $ARGUMENTS

Adapted from /novelty-check for patent legal standards. Research novelty is NOT the same as patent novelty.

Constants

• REVIEWER_MODEL = claude-cli — Model used via Claude CLI reviewer for cross-model examiner verification
• NOVELTY_STANDARD = patent — Always use legal patentability standard, not research contribution standard

Inputs

1. Invention description from $ARGUMENTS
2. patent/PRIOR_ART_REPORT.md (output of /prior-art-search)
3. patent/INVENTION_BRIEF.md if exists

Shared References

Load ../shared-references/patent-writing-principles.md for novelty/non-obviousness standards. Load ../shared-references/patent-format-us.md for 102/103 analysis framework.

Workflow

Step 1: Define Claim Elements

From the invention description, extract the key claim elements that would define the invention's scope:

1. List the technical features that make the invention novel
2. Identify which features are known from prior art vs. inventive
3. Draft preliminary claim language for 2-3 independent claims (method + system)

Step 2: Anticipation Analysis (Novelty)

For each preliminary claim, test against EACH prior art reference in PRIOR_ART_REPORT.md:

Single-reference test: Does any single reference disclose ALL claim elements?

Verdict per reference:

• ANTICIPATED: One reference discloses every element → claim is not novel
• NOT ANTICIPATED: At least one element missing from every single reference → claim is novel

Step 3: Obviousness Analysis (Inventive Step)

If the invention is novel (passes Step 2), test for obviousness:

Two/three-reference combination test: Can 2-3 references be combined to render the claim obvious?

For each combination of the top references:

1. Primary reference: Which reference is closest to the claimed invention?
2. Secondary reference(s): Which reference(s) teach the missing element(s)?
3. Motivation to combine: Would a POSITA have reason to combine these references?
   • Explicit suggestion in the references themselves?
   • Same field, same problem?
   • Common design incentive?
   • Known technique for improving similar devices?

Format as a matrix:

Step 4: Cross-Model Examiner Verification

Call REVIEWER_MODEL via claude -p with xhigh reasoning:

Write the complete fresh Claude review/help prompt to `$PROMPT_FILE`.
Preserve the role, files-to-read, objective, and required output schema from this original call shape.

If this review may need later follow-up, create and save a Claude CLI session ID on this first call.

prompt: |
    You are a senior patent examiner at the [USPTO/CNIPA/EPO].
    Examine the following invention for patentability.

    INVENTION: [invention description + preliminary claims]

    PRIOR ART: [prior art references with key teachings]

    Please analyze:
    1. Anticipation (novelty): Does any single reference anticipate any claim?
    2. Obviousness: Can any combination of references render claims obvious?
    3. Claim scope: Are the claims broad enough to be valuable?
    4. Recommended amendments if any claim is rejected.
    Be rigorous and cite specific references.

PROMPT_FILE="${PROMPT_FILE:-.aris/review-prompts/claude-review-round-N.md}"
RAW_REVIEW_JSON="${RAW_REVIEW_JSON:-.aris/review-outputs/claude-review-round-N.json}"
mkdir -p "$(dirname "$PROMPT_FILE")" "$(dirname "$RAW_REVIEW_JSON")"
CLAUDE_SESSION_ID="${CLAUDE_SESSION_ID:-$(python -c 'import uuid; print(uuid.uuid4())')}"
CLAUDE_SESSION_ID_FILE="${CLAUDE_SESSION_ID_FILE:-.aris/review-outputs/claude-session-id.txt}"
printf "%s\n" "$CLAUDE_SESSION_ID" > "$CLAUDE_SESSION_ID_FILE"
claude -p --session-id "$CLAUDE_SESSION_ID" --dangerously-skip-permissions --output-format json --model opus --effort max < "$PROMPT_FILE" | tee "$RAW_REVIEW_JSON"

Save the raw Claude CLI JSON before summarizing it. Treat the response text inside the JSON as the reviewer/helper output.

Step 5: Jurisdiction-Specific Assessment

For each target jurisdiction, provide a patentability assessment:

Under 35 USC 102/103 (US):

• Novelty: PASS / FAIL (cite specific reference if fail)
• Non-obviousness: PASS / FAIL (cite combination if fail)

Under Article 22 CN Patent Law (CN):

• 新颖性 (Novelty): 通过 / 未通过
• 创造性 (Inventive Step): 通过 / 未通过

Under Article 54/56 EPC (EP):

• Novelty: PASS / FAIL
• Inventive step: PASS / FAIL (problem-solution approach)

Step 6: Output

Write patent/NOVELTY_ASSESSMENT.md:

## Patentability Assessment

### Invention Summary
[description]

### Overall Assessment
[PATENTABLE / PATENTABLE WITH AMENDMENTS / NOT PATENTABLE]

### Anticipation Analysis
[claim-by-claim matrix against each reference]

### Obviousness Analysis
[combination analysis with motivation to combine]

### Cross-Model Examiner Review
[summary of GPT-5.5 examiner feedback]

### Recommended Claim Amendments
[If claims need modification to overcome prior art, suggest specific amendments]

### Risk Factors
[What could cause rejection during actual prosecution?]

Key Rules

• Patent novelty is absolute: any public disclosure before the priority date counts as prior art, worldwide.
• Research novelty ("has anyone published this?") is NOT the same as patent novelty ("does any single reference teach every claim element?").
• Obviousness requires BOTH: (1) a combination of references AND (2) a motivation to combine them.
• Never assume the invention is patentable just because no identical patent exists.
• The assessment is advisory only -- actual prosecution may reveal different prior art.
• If claude -p is not available, skip cross-model examiner review and note it in the output.