Works with the AI Agents you already use
Srednoff AI Agent Data Exfiltration Path Auditor
Build an evidence-linked source-to-sink path graph and prioritized egress-control/retest plan using sanitized artifacts. No traffic interception, live exfiltration probe or persistent DLP engine is supplied.
$7.99
Srednoff AI Agent Data Exfiltration Path Auditor
Example session with this skill installed
Map invoice data from local files through agent tools to remote sinks without sending it.
- Read your context and instructions
- Compiled the srednoff ai agent
Expected report shape, not a completed real audit: Executive Summary; Scope; Assets / Trust Boundaries
Connects securely to your tools. The creator never sees your data.
About this skill
Srednoff AI Agent Data Exfiltration Path Auditor
One-Line Outcome
Build an evidence-linked source-to-sink path graph and prioritized egress-control/retest plan using sanitized artifacts. No traffic interception, live exfiltration probe or persistent DLP engine is supplied.
Short Description
A network allowlist can miss model ingress, telemetry, browser uploads, URL arguments and cross-tool transfers; a reachable endpoint alone does not prove data left the boundary.
Problem It Solves
A network allowlist can miss model ingress, telemetry, browser uploads, URL arguments and cross-tool transfers; a reachable endpoint alone does not prove data left the boundary.
Data security architect: non-MCP file/customer-document-to-URL/upload/tool-argument egress graph, destination policy and residual-path register.
What You Get
- Executive Summary
- Scope
- Assets / Trust Boundaries
- Evidence Reviewed
- Threat Model
- Data Classification And Allowed Transfers
- Source-To-Sink Route Graph
- Destination / Purpose Register
- Findings
- Critical Findings
- Quick Wins
- Prioritized Remediation Plan
- Missing Evidence
- Residual Risk
- Verification Checklist
- Unknowns
- Next Validation Steps
How It Works
- Fix the review scope, revision, data classes and allowed business transfers. Read supplied sanitized artifacts only; do not transmit payloads or visit suspect endpoints.
- Inventory sources: local files, documents, conversation memory, database results, tool outputs and MCP resources. Record principal, classification and access boundary without collecting actual private content.
- Map transformations and intermediate hops: model context, summaries, tool arguments, temporary files, browser fields, clipboard and serialized logs. Encoding or compression changes representation, not authorization.
- Enumerate sinks by audience and ownership: model provider ingress, remote API, webhook, email, browser upload, navigation URL, render-fetched asset, analytics, crash telemetry, trace export and synchronized clipboard.
- Build one route per source-to-sink chain with edge evidence, actor control, principal, payload class, transfer purpose and enforcement point. Stop at missing edges and label reachability versus observed transfer.
- Review destination constraints against canonical scheme/host/port, resolved address and redirects where applicable. Separate application allowlists from actual proxy/network policy; never resolve or probe attacker-controlled URLs during review.
- Inspect URL paths/query strings, headers, tool arguments and nested metadata for data carriage; assess whether output rendering can cause an otherwise unrequested remote fetch. No request is made to demonstrate the route.
- Review cross-tool composition: a read tool plus a permitted browser or message tool may form a transfer route even if neither is individually labeled export. Track final audience and tenant on every hop.
- Determine whether each transfer is authorized for this data class, recipient and purpose. Approved model ingress is a documented transfer, not automatically a breach; TLS or a trusted domain does not authorize every payload/path.
- Review approval binding, data minimization, sink retention and failure/fallback routes. If primary export is denied but telemetry receives the payload, preserve that distinct path in the report.
- Prioritize evidenced routes and propose the smallest boundary intervention with owner and compatibility risk. Provide local inert marker/mock sink retests for allowed transfers, denied recipients, redirects and tool composition.
- Deliver graph, evidence gaps and change/retest plan. Any later egress rule or production change needs explicit scoped approval, a baseline, canary, rollback trigger and proof required legitimate transfers still work.
Best For
Security and platform engineers reviewing confidential data movement through tool-using agents
Example Tasks
- Map invoice data from local files through agent tools to remote sinks without sending it.
- Review browser uploads, URL arguments and telemetry as separate data egress paths.
- Create a safe route-based egress retest plan from sanitized architecture evidence.
Inputs
- Authorized task, data classification, permitted recipients/purposes and deployment revision
- Sanitized data-flow and tool schemas for files, model ingress, MCP resources, APIs, browser, clipboard and external writes
- Application/proxy/egress configuration and destination ownership metadata without credentials
- Sanitized trace metadata showing payload class, principal, transforms, destinations, approval and effect status
Outputs
- Executive Summary
- Scope
- Assets / Trust Boundaries
- Evidence Reviewed
- Threat Model
- Data Classification And Allowed Transfers
- Source-To-Sink Route Graph
- Destination / Purpose Register
- Findings
- Critical Findings
- Quick Wins
- Prioritized Remediation Plan
- Missing Evidence
- Residual Risk
- Verification Checklist
- Unknowns
- Next Validation Steps
Permissions
Read-only review of owner-supplied artifacts by default. Report file writes, shell/local fixture execution and official documentation lookup are optional, require available tools and their explicit scope. No live credentials are requested.
Compatibility
Standalone workflow, no Srednoff OS dependency. Codex evidence is synthetic entrypoint text only when recorded below. Claude Code, Cursor and actual buyer-runtime task delivery remain UNTESTED. Provider/platform-specific contracts must be refreshed for the installed version.
Limitations
This is an instruction/evidence workflow, not a hosted security service, persistent enforcement engine, production certification or business guarantee. The optional Python helper checks supplied claims, not underlying source artifacts. Agensi delivery is agent-only: do not save/export proprietary skill files to run local helpers; disclose NOT_EXECUTED where execution is unavailable.
Validation Transparency
Offline evidence-precheck status: PASS. Authored synthetic text-response cases: 16/16, status PASS. This is not independent security testing, complete artifact inspection or proof of deployment safety. Exact-text hashes and attempts are retained separately.
License And Publisher
Commercial License 1.0 covers original new expression; historical MIT and third-party rights remain unchanged. Mandatory marketplace conditions prevail. Publisher: Ivan Srednoff / Srednoff.Agency. No proprietary source resale or off-platform paid upsell is included.
FAQ
What is different from a generic security request? Data security architect: non-MCP file/customer-document-to-URL/upload/tool-argument egress graph, destination policy and residual-path register. No MCP or webhook transport re-audit, secret-specific lineage, data-retention policy, DLP agent, live exfiltration probe or egress enforcement.
Does it change production automatically? No. Audit/design is the default; changes, secret rotation, IAM, network probes and financial actions need separate explicit authorization.
Are the examples real incidents? No. They are synthetic DEMO DATA.
Is this a certification or guaranteed protection? No. Evidence gaps, residual risk and unexecuted tests remain explicit.
Can I use project-specific reports for clients? Yes under the included license; proprietary workflow source is not a client deliverable.
Version
1.0.0
Release Notes
Initial scoped security workflow with domain decision rules, evidence requirements, synthetic examples, regression cases and explicit authorization/runtime limits. No production/customer outcome is claimed.
This is an instruction/evidence workflow, not a hosted security service, persistent enforcement engine, production certification or business guarantee. The optional Python helper checks supplied claims, not underlying source artifacts. Agensi delivery is agent-only: do not save/export proprietary skill files to run local helpers; disclose NOT_EXECUTED where execution is unavailable.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago