KYC Rules Screener
Scores onboarding records against firm KYC/AML rules, checks documents, and routes risk.
The skill explicitly treats applicant records as untrusted documents, the rules grid as a trusted source, requires rule citations, never approves, and routes to escalation and human review. It does not explain sensitive KYC data transmission, storage, minimization, MCP permissions, user confirmation, external effects, or rollback, so points are deducted.
The steps, inputs, and JSON output are internally coherent, and every outcome is required to reference a rule. However, the rules grid and screening MCP are external dependencies, with no diagnosed failure paths for missing data, timeouts, conflicts, or malformed input, and no skill-specific test evidence; points are deducted.
The audience, invocation timing after kyc-doc-parse, inputs, and principal outputs are reasonably clear, and the skill states that it scores and routes rather than approves. Non-fit boundaries, exclusion triggers, Chinese-language support, and mainland-China network fit are not documented, so points are deducted.
The document has readable progressive steps, a table, and a JSON example. Repository context supplies an Apache-2.0 license, plugin validation, and contribution/update guidance, but the skill lacks its own version, changelog, troubleshooting guidance, stable parameter contract, and explicit maintenance owner, so points are deducted.
The skill addresses risk rating, document checks, rule-by-rule outcomes, and disposition routing in a directly usable structure for human review. The actual rules grid is not included, screening depends on an external MCP, and static review cannot verify correctness or comparative benefit; the score therefore remains at the static-review ceiling and is reduced for uncertainty.
The skill text is auditable, and repository workflows show plugin validation and secret scanning. There is no skill-specific key-path test suite, real execution record, or cross-source corroboration, so the available static evidence supports only a limited score.
- Do not treat the output as approval or a final compliance decision; verify the rules grid, screening results, and applicant record.
- Document MCP permissions, sensitive-data handling, failure feedback for missing or timed-out dependencies, human confirmation, and audit logging.
- Add key-path tests for risk rating, document status, rule citations, and disposition routing.
What does this skill do, and when should you use it?
This skill applies a firm's KYC/AML rules grid to a parsed onboarding record and screening results. It calculates a low, medium, or high risk rating and shows the factors behind it. It checks required documents for the applicant type and risk level, then reports every applicable rule with a cited rule reference. It flags missing or escalation-worthy items and does not approve onboarding.
Reads the structured record produced by kyc-doc-parse, the firm's rules grid, and sanctions, PEP, and adverse-media results from the screening MCP; calculates risk from jurisdiction, applicant type, beneficial-ownership depth, PEP exposure, screening hits, and source-of-funds clarity; checks each required document as received, missing, or expired; outputs each applicable rule with its ID, text, outcome, and evidence fields; and produces a risk rating, disposition, missing documents, escalation reasons, and rule outcomes.
- A compliance team uses it after receiving onboarding materials to score a structured applicant record against the firm's rules grid.
- A KYC analyst uses it to check which documents are required for an applicant type and risk level and to identify missing or expired items.
- A screening workflow uses it to route cases with confirmed PEP, sanctions, or adverse-media hits for enhanced due diligence.
- An operations team uses it to produce rule-by-rule outcomes and evidence fields before human review.
What are this skill's strengths and limitations?
- Covers risk rating, document checks, cited rule outcomes, and disposition routing.
- Requires a rule reference for each rule outcome, supporting reviewability.
- Explicitly treats the firm's rules grid as trusted and applicant records as derived from untrusted documents.
- Surfaces confirmed PEP, sanctions, and adverse-media hits as escalation items.
- Depends on the firm's own rules grid and a screening MCP; no generic rules or screening service is supplied.
- Requires a structured record from kyc-doc-parse first.
- The source provides no test suite, performance evidence, or standalone installation procedure.
- It never approves onboarding; final disposition remains with the escalation process and a human reviewer.
How do you install this skill?
The repository provides this skill within a plugin collection. In Cowork, open Settings → Plugins → Add plugin, paste https://github.com/anthropics/financial-services, and select the relevant agents or verticals; alternatively upload a zip containing a directory under plugins/. The README does not document a standalone installation command for this skill.
How do you use this skill?
First run kyc-doc-parse to produce a structured onboarding record. Then provide the firm's rules grid and screening MCP results, and ask the workflow to apply kyc-rules. The result should include risk_rating, disposition, missing_documents, escalation_reasons, and rule_outcomes for applicable rules. The skill scores and routes; it does not approve onboarding.