Wealth Reader Vulnerability Disclosure

generated: '2026-09-03'
method: searched
source: >-
  https://raw.githubusercontent.com/Wealth-Reader/allwr-toolkit/main/SECURITY.md ;
  https://github.com/Wealth-Reader/allwr-toolkit/security/policy ;
  probes of /.well-known/security.txt on www.wealthreader.com (404) and api.wealthreader.com (404)
program_published: true
scope_caveat: >-
  The published policy covers the allwr-toolkit open-source repository, not the Wealth Reader
  bank-aggregation API or the wealthreader.com estate. No disclosure route is published for the API
  itself: /.well-known/security.txt 404s on both hosts and no security or responsible-disclosure
  page exists on the website. For a PSD2-authorised institution holding bank credentials, that is
  the most consequential gap in this profile.
channels:
- type: private-vulnerability-reporting
  platform: GitHub
  url: https://github.com/Wealth-Reader/allwr-toolkit/security/advisories/new
  route: Repository -> Security -> Report a vulnerability
  public_issues_allowed: false
  note: Reports go directly and privately to the maintainers.
security_txt:
  served: false
  probes:
  - url: https://www.wealthreader.com/.well-known/security.txt
    status: 404
  - url: https://api.wealthreader.com/.well-known/security.txt
    status: 404
bug_bounty:
  program: none
  platforms_checked: [HackerOne, Bugcrowd, Intigriti]
  paid: false
policy:
  url: https://github.com/Wealth-Reader/allwr-toolkit/blob/main/SECURITY.md
  supported_versions: Before 1.0.0, only the latest released version receives security fixes.
  acknowledgement: The maintainers commit to acknowledging, keeping the reporter informed, and crediting them in the advisory unless they prefer otherwise.
  requested_detail: [affected version or commit, minimal reproduction, believed impact, suggested fix]
  in_scope_of_interest:
  - anything that makes the toolkit write to an unintended target
  - bypasses of the dry-run default, the plan hash verification, or the MCP write gate (ALLWR_MCP_ALLOW_WRITES)
  - credential leakage into logs, state files, reports or process arguments
  - unsafe handling of source-provided HTML or attachments
  data_handling_commitments:
  - credentials taken only from environment variables
  - secrets and personal identifiers redacted from logs and reports
  - no tokens stored in the state database
  - data sent only to the configured source and target APIs
  - no telemetry
  invitation: >-
    "If you observe behavior that contradicts any of this, treat it as a vulnerability and report
    it privately."
supporting_practice:
  ci_security_workflow: https://github.com/Wealth-Reader/allwr-toolkit/blob/main/.github/workflows/security.yml
  tooling: [bandit, pip-audit, CodeQL, dependabot]
  agent_rules: >-
    AGENTS.md forbids weakening or disabling any of those checks without an approved issue — a
    written, machine-readable governance rule aimed specifically at AI contributors.
api_side_security_controls:
  note: Observed on the API itself, for contrast with the missing disclosure route.
  observed:
  - Strict-Transport-Security on both hosts (max-age 15768000 on www, 31536000 on api)
  - X-Content-Type-Options nosniff
  - Referrer-Policy strict-origin
  - Content-Security-Policy frame-ancestors restricted to the provider's own origins
  - HMAC-SHA256 signed cards webhooks with a constant-time-comparison instruction
  - SSRF and DNS-rebinding defence re-checked before every webhook delivery
  see: security/wealth-reader-domain-security.yml