ExoTrade

ExoShield · public security model

ExoShield — the ExoTrade Integrity Defense Model

A high-assurance engineering direction built on defense in depth, least authority, measurable controls and honest evidence—not slogans.

Absolute security cannot be guaranteed. No system is unhackable or breach-proof. ExoShield is not a certification, independent audit, compliance claim, warranty or claim of protection by any named vendor.

Public model v1.1 · Last updated November 15, 2026

Security-first is an aspiration. Controls reduce risk; they do not remove it. Status labels describe repository evidence, not a live health indicator.

How to read control statuses

Current

Implemented in this repository. Production behavior still depends on correct deployment, provider configuration and acceptance testing.

Planned

A documented next step or configuration-dependent capability. It is not an active-protection claim.

External

Requires an outside provider, operator or assessor. Naming a capability does not mean ExoTrade uses it.

Independently Verified

Supported by independent assessment evidence. No ExoTrade control currently carries this status.

Global Cyber Pulse

Known exploited vulnerabilities from CISA’s Known Exploited Vulnerabilities (KEV) catalog. These are not necessarily zero-days and are not evidence that ExoTrade is affected. Applicability requires a confirmed asset inventory.

A zero-day is an unknown or unpatched vulnerability; malware and viruses are malicious software. This is public information, not ExoTrade monitoring or a SIEM system.

Ambient dots are decorative and do not indicate live protection.

FIRST EPSS estimates the daily probability of exploitation in the next 30 days. It is a prioritization signal, not a complete risk score. EPSS enrichment is not included here.

Threat feed temporarily unavailable

Last retrieved: unavailable. Last attempted (UTC): . No expired catalog is shown. Please consult the official sources above.

Back to top

Security promise

ExoTrade aims for high assurance through layered controls that can be inspected, tested and improved. A failure in one control should not automatically become a failure of every boundary. We measure progress through repeatable tests, explicit authorization rules, deployment acceptance criteria and evidence of remediation.

“Current” means a control is implemented in this repository; it does not prove every production setting or provider action. “Planned” and “External” identify work or dependencies that must not be mistaken for active defenses.

Back to top

The ExoShield seven-layer model

01

Verify Every Identity

Authenticate through Auth0, require verified-email evidence for private API access, and fail closed when identity evidence is missing or malformed. MFA, passkeys and step-up remain configuration-dependent or planned.

02

Grant Minimum Authority

Start new identities as applicants, load approved roles and staff capabilities on the server, separate back-office administration from community membership, and never trust browser-supplied authority.

03

Isolate Every Project

Check active project membership and document scope on the server. Keep public discovery limited to approved, sanitized summaries and keep private collaboration out of public responses.

04

Validate Every Input and File

Validate request schemas and identifiers, apply selective abuse limits, and use file-specific controls. Identity images are decoded and re-encoded; general documents have metadata and access controls but are not malware-scanned.

05

Preserve Tamper-Evident Evidence

Record security-relevant actions with redacted metadata, use revision checks for sensitive decisions, and preserve issued signing artifacts and published governance records. These measures are not a claim of an indestructible or independently verified log.

06

Detect, Contain, and Respond

Use application audit events, bounded errors and a documented incident playbook as a foundation. Central monitoring, managed detection, on-call response and exercised containment require further operational work.

07

Recover, Test, and Improve

Run security regression, authorization, privacy and failure-path tests alongside type and production-build checks. Backup restoration, disaster-recovery exercises and independent testing remain release gates, not current claims.

Back to top

Current protections

These controls are implemented in source. Their presence does not replace live deployment acceptance or independent assessment.

Informational threat-feed safeguards

Current

Public CISA catalog retrieval is bounded, validated, cached and rendered as text with an unavailable fallback. This is informational only, not monitoring, a SIEM system or evidence that ExoTrade is affected.

Auth0 token verification

Current

Private APIs verify Auth0 access tokens with the expected issuer, audience, algorithm and expiry. Missing identity-provider configuration fails closed; provider claim configuration still requires live acceptance.

Verified-email requirement

Current

Private API access requires a verified-email claim. Missing, malformed or unverified identity evidence is rejected rather than treated as authenticated.

Server-side authorization

Current

Approved roles and staff capabilities are loaded from server-side account records. New identities are applicants by default, and browser-provided roles or user IDs do not grant authority.

Role and admin boundaries

Current

Back-office administration is distinct from community participation. Limited staff remain capability-scoped, while administrative access does not automatically create project membership.

Project and document isolation

Current

Project APIs check membership and document scope on the server. Public project discovery uses approved, sanitized summaries rather than private project records.

Trusted-server data boundary

Current

Database migrations deny direct anonymous and ordinary authenticated table access and enable row-level security. Private reads and writes pass through authorized server code; applied production settings still need verification.

Request and schema validation

Current

Request bodies use schemas, route identifiers are validated, and JSON routes enforce actual streamed-byte limits rather than trusting a declared size. This does not eliminate every possible malformed-input risk.

Selective application-level throttling

Current

Some sensitive or public actions have application-level limits. This is not universal, atomic, edge-based rate limiting and does not replace denial-of-service protection.

Identity-image processing

Current

Identity uploads are limited to JPEG or PNG, checked for type, size and dimensions, decoded and re-encoded to remove metadata, and stored privately with random keys. This is not antivirus scanning.

Project-file controls

Current

Project documents use extension, declared-type and size checks, sanitized filenames, random storage keys, scoped authorization and short-lived download links. Uploaded document contents are not malware-scanned.

Integration-specific outbound controls

Current

The TelePath integration restricts outbound requests to a fixed HTTPS destination with bounded fetching and redirect rejection. This is not a universal SSRF-control claim for every possible integration.

Audit and revision evidence

Current

Security-relevant workflows record redacted audit metadata; sensitive decisions use revision checks, and issued signing artifacts and published governance records have immutable or append-only behavior.

Explicit API authorization

Current

Private API requests use explicit bearer authorization rather than ambient session cookies. Conventional cookie CSRF is therefore not the primary boundary; XSS and browser-token protection remain important.

Security regression tests

Current

The repository includes tests for identity claims, authorization, project isolation, privacy, uploads, stale decisions and failure paths. Local execution is verified; enforced third-party CI gates are not.

Back to top

Planned, external and independently verified

Planned or configuration-dependent

MFA and step-up authentication

Planned

Optional server policy support exists, but MFA is currently default-off and no enabled tenant policy has been verified. Passkeys and broader step-up coverage are future work.

Backups and tested restoration

Planned

No backup schedule, completed restore test, recovery-time objective or recovery-point objective is evidenced here. Recovery readiness must be demonstrated before being claimed.

Central monitoring and alerting

Planned

Application audit and error records exist, but centralized security alerting, exercised escalation and continuous detection coverage are not established.

Dependency, secrets and SAST automation

Planned

Dependency scanning, SBOM generation, secrets scanning, security-focused static analysis and enforced security release gates are not evidenced in repository automation.

Documented peer-review and branch gates

Planned

Local tests and builds are current verification steps. Mandatory peer review, protected branches and externally enforced release policies require separate evidence.

Bug bounty

Planned

A later bounty may be considered after scope, safe-harbor terms, triage capacity and funding are established. No bounty is offered now.

External dependencies

Encryption and browser headers

External

HTTPS certificates, HSTS and other browser security headers depend on hosting and edge configuration. Their live settings have not been independently verified in this review.

Global CDN, WAF, DDoS and bot defense

External

Distributed edge capacity and bot intelligence require an external network provider and correctly configured deployment. No provider is claimed as active.

Managed SIEM and on-call SOC

External

Round-the-clock monitoring, correlation and response require operational people and services beyond application code.

Mature malware scanning and content disarm

External

General uploaded files are not malware-scanned. A mature scanning, sandboxing or content-disarm pipeline requires specialized services and safe quarantine design.

Independent penetration testing

External

An independent application and deployment assessment must be completed before sensitive or high-value activity. This source review is not a penetration test.

Independently Verified

No controls are independently verified in this release. No independent penetration test, security audit, certification or vendor-protection verification has been evidenced. The independent-verification register is intentionally empty.

Back to top

Identity and access

  • Current: Auth0-based authentication, verified-email enforcement, server-side token verification and database-backed authorization.
  • Current: Applicant-by-default access, separately approved community roles, project membership checks and distinct staff/admin boundaries.
  • Planned or configuration-dependent: MFA, passkeys and step-up authentication. Optional policy support is not evidence that a tenant has enabled or tested them.
  • Review principle: Administrative authority does not automatically confer community membership, and user-interface visibility is never treated as authorization.
Back to top

Application and API controls

Request schemas, identifier validation and server authorization are current. Selected workflows have application-level throttles, but ExoTrade does not claim universal or edge rate limiting.

Files are handled by class

Identity images receive specialized decoding and re-encoding. General project documents receive declared-metadata, size, name, authorization and scoped-link controls. Neither statement means uploaded files are malware-scanned.

Outbound requests need explicit boundaries

The current TelePath integration has a fixed-destination HTTPS policy, bounded fetch and redirect rejection. Future integrations need their own SSRF review rather than inheriting this claim automatically.

Browser and request context

Private APIs use explicit bearer authorization instead of ambient session cookies, so conventional cookie CSRF is not the primary boundary. Secure cookies, HSTS, CSP and related response headers remain hosting/configuration concerns where applicable; no strict-CSP claim is made.

Back to top

Data and project isolation

  • Project membership, document scope and back-office permissions are checked by server code. Public summaries are sanitized and separately approved.
  • Database migrations enable row-level security and remove direct public table access. Privileged server access still requires correct authorization on every operation; RLS alone is not the whole isolation model.
  • Least-privilege service access and private storage boundaries reduce exposure, but applied production policies and storage settings require deployment verification.
  • External/configuration-dependent: HTTPS and encryption-in-transit settings are supplied by the hosting and provider path. Their live state is not independently verified here.
  • Planned: Backup schedules and restore tests are not evidenced. Do not infer recovery readiness from the existence of a database.
Back to top

Secure development lifecycle

The near-term verification target is OWASP ASVS 5.0 Level 2, with selected Level 3 controls for admin, contract, future financial, identity and sensitive-document workflows. This is a mapping target, not an ASVS certification or independent assessment.

Review the official OWASP ASVS 5.0 standard
Secure-development control register
ControlStatusEvidence or gap
Threat modelingCurrentA high-level trust-boundary, asset, threat and mitigation model is documented. It must be revisited as features and deployment change.
Peer reviewPlannedA mandatory, documented review policy and branch protection are not evidenced.
Automated testsCurrentUnit and regression tests cover security-sensitive rules; the full suite, TypeScript and production build are local release checks.
Dependency scanningPlannedNo repeatable repository scanning workflow or reviewed vulnerability baseline is evidenced.
Software bill of materialsPlannedA versioned SBOM should be generated and retained for release candidates.
Secrets scanningPlannedAutomated history and change scanning is not evidenced. Public configuration must remain distinguishable from credentials.
Security static analysisPlannedType checking and linting are useful but are not a substitute for security-focused SAST and reviewed findings.
Enforced release gatesPlannedLocal checks are current; externally enforced CI, approval and deployment policies require separate configuration evidence.
Back to top

Monitoring, response and recovery

Current: security-relevant application events and redacted audit metadata provide investigation evidence, unexpected API failures return generic responses, and a minimum incident playbook is documented for maintainers.

Planned: centralized alerting, exercised escalation, managed detection, on-call coverage, recovery objectives and repeated restore exercises. No 24/7 monitoring or recovery-time claim is made.

Response priorities are to protect people, preserve evidence, contain unauthorized access, communicate without speculation, recover deliberately and document corrective action. Reporters and participants can use the existing help-request form; the Help hub explains ordinary account and access issues.

Back to top

What requires an external service

Application code can enforce identity, authorization and validation, but it cannot credibly recreate global network capacity, independent human review or a staffed security operation by itself.

Global edge CDN, WAF, DDoS and bot intelligence

Distributed networks absorb and filter traffic before it reaches the application. Effective bot defense also requires broad telemetry and continuously tuned detection.

Managed SIEM and on-call SOC

Central correlation, trained responders and continuous escalation require an operating team and tooling beyond a web application.

Mature malware scanning and content disarm

Safe detonation, signature and heuristic coverage, quarantine and document reconstruction require specialized, maintained services.

Independent penetration testing

A credible independent assessment requires a qualified assessor with documented scope, methodology, findings and remediation review.

A later bug bounty

A bounty needs explicit scope, safe-harbor terms, triage capacity and funding. It is not launched by publishing a reporting form.

Back to top

Vendor-neutral enterprise comparison

Cloudflare, Akamai and AWS are comparison examples only; their inclusion does not assert current use, endorsement or an affiliate relationship. Enterprise pricing is generally quote-based. AWS Shield Advanced publishes a USD 3,000/month base subscription, with a one-year commitment plus usage and support considerations; confirm current terms before budgeting. This is information, not a recommendation to purchase it.

Capability questions for an enterprise evaluation
Evaluation areaCloudflare EnterpriseAkamaiAWS-native edge
Typical building blocksCDN, DNS, WAF, DDoS and bot products; exact features depend on contract.Enterprise delivery, application security and mitigation products; exact bundle depends on contract.CloudFront, WAF, Shield and related services; architecture and support choices affect coverage.
Evaluation fitConsider when consolidated edge DNS, application delivery and security operations are priorities.Consider for complex enterprise delivery, existing Akamai operations or specialized service requirements.Consider for AWS-centered workloads and teams able to operate integrated AWS controls.
Commercial approachEnterprise terms are generally custom; request a written quote and service scope.Enterprise terms are generally custom; request a written quote and service scope.Shield Advanced has a public base subscription; usage, support, commitment and other service charges need review.
What to testOrigin lock-down, rule tuning, logs, false positives, escalation and portability.Migration effort, rule operations, logs, service response, integration and portability.Origin exposure, WAF coverage, support tier, logs, cost drivers and cross-service operations.
Back to top

Security roadmap

  1. Phase 1 · Foundation now

    Prove the application boundary

    Maintain verified identity, least authority, project isolation, validation, audit evidence and regression tests. Document gaps honestly and exercise the incident playbook.

  2. Phase 2 · Before broad beta

    Add edge and operational protection

    Evaluate and configure managed CDN/WAF/DDoS and bot controls, security headers, central alerting, backup restoration and vulnerability-management automation. Acceptance requires deployment evidence and negative tests.

  3. Phase 3 · Before sensitive or high-value activity

    Obtain independent verification

    Complete scoped penetration testing, remediate findings, verify identity and document workflows, and rehearse incident and recovery procedures. Selected higher-assurance controls should cover admin, contract, identity and sensitive-document paths.

  4. Phase 4 · When money flows

    Continuous assurance

    Require stronger authentication and step-up, transaction-specific threat modeling, continuous monitoring, repeatable recovery exercises, dependency/SBOM governance and periodic independent reassessment before enabling financial activity.

Phase 1 remains nonfinancial. Later phases are entry gates, not promises that a date, vendor, assessment or financial feature has already been scheduled.

Back to top

Report a vulnerability

Use the existing security and help contact path to report a suspected vulnerability. Include the affected public page or workflow, a minimal reproduction, expected and observed behavior, and potential impact. Do not include passwords, tokens, identity documents or another person’s data in the initial report.

  • Do not access, change, download or retain another person’s data.
  • Do not disrupt service, degrade availability, use social engineering or run high-volume automated testing.
  • Stop at the minimum proof needed, preserve confidentiality and give maintainers an opportunity to investigate and remediate.
  • For ordinary account or feature questions, start with the Help hub.

No bug bounty is offered yet. Reporting does not create a reward, employment, legal or confidentiality agreement.

Open the help-request form
Back to top
Security-first: read the ExoShield model