claim-extraction
Use when a report or article needs to be split into individual claims before it is graded, cited or used as evidence — the user asks "what does this report actually claim?", "split this into claims", "separate the facts from the judgements", or another skill needs a claim list. Turns one document into five to twelve atomic claims, each typed (observation, attribution, assessment, actor claim, victim disclosure) and anchored to the verbatim sentence it rests on, so every extraction can be checked against the source. Preserves the source's own hedging. Does not grade. Invoked by /quality-of-information-check; its output feeds /ach, /key-assumptions-check, /intelligence-writing, /writing-assessments and /threat-actor-profiling.
Liberty91LTD · 已有静态检查记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add Liberty91LTD/cti-skills --skill claim-extraction技能说明
name: claim-extraction description: Use when a report or article needs to be split into individual claims before it is graded, cited or used as evidence — the user asks "what does this report actually claim?", "split this into claims", "separate the facts from the judgements", or another skill needs a claim list. Turns one document into five to twelve atomic claims, each typed (observation, attribution, assessment, actor claim, victim disclosure) and anchored to the verbatim sentence it rests on, so every extraction can be checked against the source. Preserves the source's own hedging. Does not grade. Invoked by /quality-of-information-check; its output feeds /ach, /key-assumptions-check, /intelligence-writing, /writing-assessments and /threat-actor-profiling. user-invocable: true metadata: version: 1.0.0 tags: [tradecraft, claims, sourcing, sat] tradecraft: true
claim-extraction
An event is a bundle of claims with very different evidentiary standing. "Akira exploited a SonicWall zero-day to hit a US hospital and exfiltrated 400GB" is at least four claims:
| Claim | Type | Where it comes from |
|---|---|---|
| The hospital suffered an intrusion | victim_disclosure | The hospital's own notification |
| Initial access was through a SonicWall vulnerability | observation | A vendor's incident response finding |
| The intrusion was carried out by Akira | attribution | A leak-site posting plus TTP overlap |
| 400GB of data was taken | actor_claim | The actor's own statement |
One grade on the event either averages these into something meaningless or takes the strongest and inflates the weakest. This skill splits the document so each claim can be graded on its own.
Composition. Uses the fetcher in /source-provenance. Invoked by /quality-of-information-check. Output is consumed by /ach, /key-assumptions-check, /intelligence-writing, /writing-assessments, /threat-actor-profiling and /campaign-tracking.
Inputs
- Document text, or a URL.
- Optional focus: "attribution only", "IOCs and TTPs only", or the question the claims are meant to serve.
- Optional: the provenance result from
/source-provenance, so that article-only material can be told apart from the primary's claims.
For a URL, get the text through the provenance fetcher, which honours robots.txt and keeps the full text in a local cache:
python3 skills/source-provenance/scripts/resolve_provenance.py text <url>
It prints the path to a local text file. Read that file. If the fetch fails, ask the user to paste the text.
Output
A claim table, and a short note on what was deliberately left out.
**Document:** <title> (<organisation>, <date>)
**URL:** <url>
**Claims extracted:** 8
| ID | Claim | Type | Anchor (verbatim) | Location |
|---|---|---|---|---|
| C01 | ... | observation | "..." | paragraph 3 |
**Not extracted:** background on the actor's 2023 activity (restated from earlier
reporting, not load-bearing); product recommendations; mitigation guidance.
And the same as JSON, matching the claim_id, claim, claim_type and anchor fields of the evidence item schema in /quality-of-information-check:
{
"document": {"url": "https://...", "title": "...", "org": "...", "date_published": "2026-09-25"},
"claims": [
{
"claim_id": "C01",
"claim": "The actor used a WAF bypass to reach exposed PeopleSoft servers.",
"claim_type": "observation",
"anchor": {
"document_url": "https://...",
"sentence": "verbatim sentence from the document",
"location": "paragraph 3"
},
"date_observed": "2026-09-18",
"flags": []
}
],
"not_extracted": ["..."]
}
date_observed is when the activity happened, if the document says. Otherwise "unknown". It is not the publication date.
Procedure
1. Read the document once
Identify the bottom-line statements and the facts they rest on. Do not start writing claims until you have read to the end: the caveats are often in the last section.
2. Write each claim
For each candidate:
- Rewrite it as one atomic sentence. One subject, one assertion, no compound "and".
- Tag the type (rules below).
- Copy the anchor sentence or sentences verbatim. Copy, do not retype from memory. Every character must be findable in the source.
- Record the location: paragraph number or the heading it sits under.
3. Type rules
| Type | It is | Examples |
|---|---|---|
observation | Something seen or measured | A hash, an IP address, a technique executed, the date of activity, a count from telemetry |
attribution | Linking activity to an actor, a country or a campaign | "We track this as UNC1234", "linked to the MSS" |
assessment | A judgement about intent, capability, future behaviour or significance | "The group is likely to expand to OT targets" |
actor_claim | A statement that originates with the threat actor | Leak-site posting, ransom note, forum post, statement to a journalist |
victim_disclosure | A statement from the affected organisation, or its regulatory filing | 8-K, breach notification letter, press statement |
When a claim could be two types, pick by asking what would have to be true for the source to know it:
- If the source would have to have seen it:
observation. - If the source would have to have reasoned to it:
attributionorassessment. - If the source is repeating what the actor or victim said:
actor_claimorvictim_disclosure, whoever is relaying it.
The boundary cases are worked through in references/claim-typing-guide.md.
4. Preserve the hedge
Where the document hedges, the hedge is part of the claim.
| Source says | Claim reads | Not |
|---|---|---|
| "We assess with moderate confidence that the activity is linked to APT41" | "Vendor assesses with moderate confidence that the activity is linked to APT41." | "The activity is li |