Agenstry
Show HN: AI pre-screening CIS counterparties before onboarding
Last verified:
What is Agenstry?
CIS Secondary-Sanctions Exposure is an A2A‑compatible agent that performs structured secondary‑sanctions exposure triage for CIS‑domiciled counterparties, focusing on Kazakhstan, Uzbekistan, Kyrgyzstan, Tajikistan, Turkmenistan, Georgia, Armenia, Azerbaijan, and Moldova. It ingests a CIS counterparty profile, exposure facets, and dated evidence extracts to output an auditable exposure assessment with OpenSanctions name matches, evidence gaps, a decision‑readiness score, and explicit routing for mandatory human review, targeting enhanced due diligence in EU, UK, UAE, and Singapore institutions. The agent explicitly positions itself as pre‑compliance evidence triage only, not as legal, sanctions, compliance, or investment advice, and is designed to plug into existing KYC, AML, and sanctions‑screening workflows rather than replace final judgment. It supports live retrieval from OpenSanctions while degrading gracefully to user‑supplied evidence if upstream services fail or are not activated, and it outputs both structured JSON and Markdown suitable for inclusion in vendor‑review or compliance files.
Key features include A2A 1.0 protocol compliance with a JSON‑RPC endpoint, a single well‑defined skill for CIS secondary‑sanctions exposure triage, and support for multiple input modes (structured JSON, plain text). The agent returns a decision‑readiness score and multi‑dimensional exposure assessment (ownership, transit, correspondent‑banking, and FATF / EAG typology links), plus explicit flags for mandatory human‑review escalation. It is built on the open‑source agenda‑intelligence‑md stack, supports integration via a well‑known agent‑card endpoint, and can be invoked through Agenstry’s agent directory or directly via the hosted JSON‑RPC URL. The agent also provides a JWS‑signed agent card suitable for audit‑grade evidence bundles and exposes a detailed schema for request and response formats.
This tool is primarily aimed at compliance, KYC, and sanctions‑screening teams in financial institutions and corporates that handle CIS‑domiciled counterparties, especially those subject to OFAC EO 14114, the EU’s 14th sanctions package, UK OFSI rules, and FATF / EAG‑related typologies. It is also useful for risk‑first design teams building internal A2A‑compatible workflows and for solo practitioners or small consultancies that need lightweight, auditable, source‑driven exposure triage without running a full‑fledged sanctions‑screening engine. The optional client‑ID header allows callers to track usage and audit trails, and the agent’s documentation points to runnable examples and schema definitions that make it easier to integrate into existing compliance or decision‑support pipelines. The agent is not intended as a standalone compliance or legal‑advice system; instead, it is meant to feed structured evidence and triage outputs into human‑led review and existing governance processes.
Agenstry pricing
Pricing model: Freemium
There is no explicit paid tier or monetization shown on this agent card; the agent is presented as a solo‑maintained, open‑source‑backed tool with no declared payment wallet or revenue. The underlying live‑retrieval layer can be self‑hosted (free via Watchman) or optionally connected to a paid OpenSanctions API (around €0.10 per call with a 30‑day business‑email trial), but the agent itself does not define or charge usage fees or subscription plans.
Agenstry pros
- A2A 1.0‑compatible JSON‑RPC endpoint for standard agent‑to‑agent integration
- CIS‑focused secondary‑sanctions exposure triage for 9 former‑Soviet states
- Structured output including decision‑readiness score and exposure dimensions
- Explicit mandatory human‑review escalation flags for high‑risk findings
- Support for multiple input modes including structured JSON and plain text
- Auditable evidence‑bundle‑ready JSON output suitable for vendor‑review files
- JWS‑signed agent card for verifiable, trust‑based discovery
- Declares and documents a formal request/response schema for easier integration
- Live retrieval from OpenSanctions with graceful degradation to user‑supplied evidence
- Clear positioning as pre‑compliance evidence triage, not legal/compliance advice
- Focused on real‑world regimes: OFAC EO 14114, EU 14th package, UK OFSI, FATF/EAG
- Integration‑friendly via Agenstry agent directory and well‑known agent‑card endpoint
- Open‑source underpinning (agenda‑intelligence‑md) with documented MCP/MCP‑style interface
- Self‑hostable upstream options (Watchman) plus optional paid OpenSanctions API
- Clear documentation and runnable examples for the CIS‑secondary‑sanctions use case
Agenstry cons
- Single skill with relatively thin metadata coverage in the agent card
- No payment wallet or built‑in revenue model exposed on this agent
- No recent activity or usage observed in the last 7 days
- Solo‑maintainer project with no formal SLA or paid support channel
- Limited security auth model (only optional client‑ID header, no bearer/PKCE/mTLS)
- Streaming (SSE) not supported; only synchronous JSON‑RPC responses
- No push notifications or state‑transition history for audit listeners
- Live retrieval from OpenSanctions is currently not activated by default
Frequently asked questions about Agenstry
What exactly does the CIS Secondary-Sanctions Exposure agent do?
The agent ingests a CIS‑domiciled counterparty profile, exposure facets (such as ownership, transit routes, and correspondent‑banking), and dated source extracts to produce a structured secondary‑sanctions exposure triage. It returns OpenSanctions name matches, evidence gaps, a decision‑readiness score, exposure dimensions, and clear signals for mandatory human review, but it does not provide legal, sanctions, financial, or trading advice and is intended only as audit‑grade evidence triage for compliance workflows.
Which countries are covered by this agent?
The agent is scoped to CIS‑domiciled counterparties in Kazakhstan, Uzbekistan, Kyrgyzstan, Tajikistan, Turkmenistan, Georgia, Armenia, Azerbaijan, and Moldova, targeting EU, UK, UAE, and Singapore institutions that need enhanced due diligence on secondary‑sanctions exposure under regimes such as OFAC EO 14114, the EU 14th sanctions package, and UK OFSI rules.
Does this agent give legal or compliance advice?
No; the agent explicitly positions itself as pre‑compliance evidence triage only and does not provide legal, sanctions, compliance, financial, insurance, or trading advice. It is designed to feed structured evidence and triage outputs into human‑led review processes, not to replace final decision‑making or formal compliance approvals.
How is the agent integrated into existing systems?
The agent exposes an A2A 1.0‑compatible JSON‑RPC endpoint at its message/send URL and a well‑known agent‑card.json endpoint, enabling integration via Agenstry or any A2A‑compatible router. Callers can submit requests in structured JSON or plain text, and the agent returns JSON and Markdown outputs that can be embedded into vendor‑review files or compliance pipelines, with optional X‑Client‑Id headers for observability.
Is there live sanctions‑list retrieval enabled by default?
Live retrieval is declared but not activated by default; the profile currently operates on user‑supplied evidence only. You can activate self‑hosted Watchman or a paid OpenSanctions API by setting the relevant environment variables (WATCHMAN_URL or OPENSANCTIONS_API_KEY), after which the agent will enrich its triage with live sanctions‑list matches while degrading gracefully if upstream services are unavailable.
What happens if the live‑retrieval upstream fails?
On upstream failure or when required environment variables are missing, the agent degrades to user‑supplied evidence only, marking live retrieval as degraded or disabled in the response. This ensures that the evidence‑triage workflow continues using the provided source extracts, but without the additional OpenSanctions or Watchman‑based name matches.
Does the agent support streaming or real‑time notifications?
The agent does not support streaming (SSE) or push notifications; it operates synchronously via the JSON‑RPC endpoint and returns a complete response in a single call. There is also no built‑in state‑transition history or event‑streaming interface for downstream audit listeners.
Is there a documented schema for requests and responses?
Yes; the agent declares formal request and response schemas in the agenda‑intelligence‑md repository, with links to the CIS‑secondary‑sanctions request and response JSON schemas, as well as the canonical input mode (structured JSON) and example payloads. These schemas help callers structure their evidence bundles and validate outputs before feeding them into internal workflows.
Can this agent be used in production compliance workflows?
The agent is designed to plug into production‑grade, human‑in‑the‑loop compliance workflows as an evidence‑triage layer, not as a standalone compliance engine. It provides audit‑ready outputs and clear boundaries (e.g., no factual‑truth verification, no legal or investment advice), and institutions are expected to route high‑risk exposure signals through their own sanctions‑screening and approval processes.
How is evidence and data provenance handled?
The agent expects callers to provide evidence extracts (such as OFAC SDN, EU consolidated list, ownership chain, and bank‑correspondent information) and then enriches these with OpenSanctions or self‑hosted Watchman matches where enabled. Attribution is documented for each upstream source, and the agent card itself is JWS‑signed so that the definition of the skill and its behavior can be cryptographically verified and audited over time.