Vendor Security Assessment

Security Whitepaper — Vendor Assessment for Procurement Teams

A structured security and compliance assessment for procurement, InfoSec, and GRC teams evaluating Compliance Glossary for Confluence. Production v6 runs on Atlassian Forge with no DailyMind-operated app servers. Its only external egress is an optional, administrator-enabled GitHub connector that calls https://api.github.com. AI-assisted curation uses Atlassian-hosted Forge LLM and starts only when an administrator selects findings. See the privacy policy. (Scope note: this marketing site, hosted on Cloudflare Pages, uses self-hosted, cookieless analytics — Umami and GoatCounter on vigilcom.dev — and no third-party trackers; the app itself does not.)

Compliance Glossary is a Forge app for Confluence, published by DailyMind LTD in Cyprus. Production v6 stores core records in Forge KVS and enterprise evidence in Forge SQL and Object Store. Its only external egress is an optional, administrator-enabled GitHub connector. DailyMind runs no app servers and has no access to customer data at rest. Atlassian maintains the SOC 2 and ISO 27001 certifications for that infrastructure.

1. Executive Summary

Compliance Glossary for Confluence is a terminology governance application built on Atlassian Forge — Atlassian’s serverless app platform. In production v6, code execution, customer-data storage, and processing occur within Forge, with no DailyMind-operated app server or database. The only external egress is an optional GitHub connector that calls https://api.github.com; it is off until an administrator enables it with a customer-provided GitHub App installation. Because of this egress, v6 is not eligible for Runs on Atlassian. AI-assisted curation (since v5.0) uses Atlassian-hosted Forge LLM, starts only when an administrator selects findings, and creates drafts only; see the AI-assisted curation documentation and the privacy policy. (Scope note: this marketing site, hosted on Cloudflare Pages, uses self-hosted, cookieless analytics — Umami and GoatCounter on vigilcom.dev — and no third-party trackers. The Forge app itself does not load these.)

The vendor is DailyMind LTD (Registration: HE 439959), registered in Cyprus (European Union). DailyMind publishes the application on the Atlassian Marketplace and provides support, but does not host, process, or have access to customer data at rest.

This whitepaper documents the application’s architecture, data handling practices, security controls, and compliance framework alignment to support vendor risk assessments, procurement due diligence, and third-party security reviews.

2. Architecture Overview

The application runs within the Atlassian Forge platform; the optional GitHub connector is its only external egress. The following diagram shows the data flow and isolation boundaries:

UserConfluence UI
→
Forge SandboxNode.js 22.x
Isolated per installation
→
Forge StorageKVS, SQL, Object Store
Per-installation data
BlockedNo DailyMind servers
×
RestrictedOutbound only to api.github.com
(optional, admin-enabled)
×
BlockedNo third-party databases
Forge runtime constraints (app scope only): The Forge sandbox restricts network access, file system access, and available APIs. The Compliance Glossary Forge app can make backend HTTP requests only to https://api.github.com, the single address declared in its manifest for the optional GitHub connector. It cannot access the file system or spawn processes. Apart from GitHub connector requests, customer terminology data remains inside the customer’s Atlassian Forge environment; Forge LLM is Atlassian-hosted.

Key Architecture Properties

3. Data Residency

Core records are stored in Forge KVS (Entity Store); enterprise evidence uses Forge SQL and Forge Object Store. Storage location is set by Atlassian’s Forge hosting, not by DailyMind. Production v6 does not claim Atlassian Marketplace data residency support.

4. Authentication & Access Control

The application delegates all authentication to the Atlassian platform and enforces role-based access control with a mandatory four-eyes approval workflow.

Audit note: The four-eyes enforcement is a technical control, not a policy control. The system rejects self-approval at the API level regardless of the user’s role or permissions. Every approval action records the submitter, reviewer, timestamp, and decision rationale in an append-only audit history table stored in Forge Key-Value Store / Entity Store, with server-enforced write restrictions (no delete or modify API is exposed) and Atlassian platform-level write timestamps. This audit history is append-only at the application layer; it is not cryptographically chained or hash-sealed — it does not provide cryptographic tamper-evidence. Tamper resistance derives from the absence of a delete/modify code path and from the underlying Forge storage controls operated by Atlassian, not from cryptographic verification.
On electronic signatures: The approval workflow records who approved what and when, with a decision rationale. This is an approval workflow with audit trail, not a 21 CFR Part 11-compliant electronic signature implementation. It does not include the signature manifestations required by §11.50 (printed name of signer, date/time, and meaning bound to the signed record in human-readable form) or the two-component identification, non-repudiation, and component-control requirements of §11.200. Customers in FDA-regulated environments operating under Part 11 must layer a Part 11-compliant electronic signature system on top.

5. Data Processing Inventory

The following table documents all data categories, whether they are stored, their purpose, and retention period:

Data Category Stored Purpose Retention
Glossary term definitions Yes Core functionality — managed terminology Until deletion by customer
Audit trail (account IDs) Yes Regulatory compliance — who did what, when Until deletion by customer
Scan results Yes Compliance evidence — page scan findings Until deletion by customer
Page content Baseline freeze only Scans process page content in memory. A baseline freeze stores the exact scanned revision as evidence in Forge Object Store See privacy policy
AI curation proposals and usage verdicts Yes Proposal metadata and page references; usage verdicts with model identifiers and prompt versions. Prompts, excerpts, and raw model responses are not stored See privacy policy
GitHub App private key and webhook secret (optional connector) Yes, Forge Secure Storage Authenticate the customer’s own GitHub App Deleted when the connector is disabled
User credentials No Delegated to Atlassian identity provider N/A
Email addresses No Not collected or processed N/A
IP addresses No Not collected or processed N/A
Cookies No Application uses no cookies, local storage, or client-side tracking N/A
Minimal data footprint: The application stores only what is necessary for core functionality and regulatory audit trail requirements. No personal data beyond Atlassian account IDs is stored. No marketing data, behavioral analytics, or usage telemetry is collected.

6. Sub-Processors

Production v6 uses Atlassian as its sub-processor for app data, including Atlassian-hosted Forge LLM. GitHub is involved only when an administrator enables the connector with the customer’s own GitHub App. No analytics providers, CDNs, or DailyMind-operated infrastructure process app data.

Sub-Processor Location Role Data Processing Agreement
Atlassian Pty Ltd Sydney, Australia (global infrastructure) Infrastructure, compute (Forge runtime), storage (KVS, SQL, Object Store), Atlassian-hosted Forge LLM (AI-assisted curation, approved-term usage classification) Atlassian’s standard DPA
GitHub (customer-enabled connector only) Customer’s GitHub account Repository reads through the customer’s own GitHub App installation, only after an administrator enables the connector Customer’s own GitHub terms

No other sub-processors are used by production v6. It does not integrate with analytics platforms, error tracking services, or content delivery networks for app data. Data flows for GitHub and Forge LLM are described in the privacy policy (§3.4–3.5) and our Data Processing Agreement.

Marketing-website processors (Cloudflare, Inc., Anthropic PBC, Resend, Inc.) are documented in privacy.html; they do not process App data within Atlassian Forge.

7. Encryption

All encryption is managed by the Atlassian platform. The application does not implement its own cryptographic operations.

Layer Standard Managed By
At rest AES-256 Atlassian infrastructure
In transit TLS 1.2+ Atlassian infrastructure
Key management Atlassian KMS Atlassian infrastructure

8. Incident Response

DailyMind maintains a structured incident response process for security events affecting the application:

  1. Detection — security incidents are detected through Atlassian platform monitoring, automated testing, user reports, or vulnerability disclosures to [email protected].
  2. Assessment — initial triage and impact assessment within 4 hours of detection. Classification of severity (critical, high, medium, low) and identification of affected installations.
  3. Containment — immediate containment actions including, where applicable: emergency patch deployment, feature disablement, or coordinated response with Atlassian’s security team.
  4. Notification — affected customer (the controller) notified without undue delay per GDPR Article 33(2), and in any case within 72 hours where feasible, to enable the controller to meet its own Article 33(1) notification deadline. Notification includes: description of the incident, data categories affected, likely consequences, and remediation measures taken.
  5. Post-incident review — root cause analysis, remediation verification, process improvement recommendations, and post-incident summary published to affected customers.

Security contact: [email protected]

Vulnerability disclosure: we acknowledge reports within 3 business days and provide an initial assessment within 10 business days. We do not take legal action against researchers who report vulnerabilities in good faith.

For incidents related to the underlying Atlassian Forge platform or infrastructure, Atlassian’s own incident response process applies.

9. Business Continuity

The application’s business continuity posture is strengthened by its dependency on Atlassian’s infrastructure rather than vendor-operated systems.

Source Code & Development

Data & Storage

Vendor Continuity

10. Vulnerability Management

The application follows a multi-layered vulnerability management approach combining Atlassian’s platform security with application-level controls.

Control Description
Atlassian Forge security review All Forge apps undergo Atlassian’s security review before marketplace listing, including static analysis and permission scope review.
npm audit gate Zero known vulnerabilities policy. npm audit --omit=dev runs before every deployment. Any vulnerability blocks the deploy.
Automated test suite 77 automated tests covering core functionality, access controls, four-eyes enforcement, and audit trail integrity.
Static analysis ESLint with strict rules enforced via pre-commit hooks. No code merges without passing lint.
Dependency management Regular dependency updates. Security advisories monitored for all direct and transitive dependencies.
Atlassian critical fix deadline Atlassian requires marketplace apps to fix critical vulnerabilities within 10 days of notification.

11. Compliance Framework Mapping

The following table maps application capabilities to specific requirements across regulatory and industry frameworks:

Framework Requirement How Addressed
GDPR Art. 28 (source) Processor DPA published, minimal PII (account IDs in the audit trail; customer-supplied content per the DPA), right to erasure automated via Atlassian’s Personal Data Reporting API
SOC 2 CC6.1 Logical access controls Forge authentication (asApp()/asUser()), role-based access, four-eyes approval workflow
ISO 27001 A.5.3 Segregation of duties Four-eyes approval workflow enforced server-side; submitter cannot approve own terms
DORA Art. 6 (source) ICT risk management framework Documented terminology governance, full audit trail, change tracking with attribution
DORA Art. 8 (source) Identification Version history on every term, complete change tracking, exportable audit records supporting classification and documentation of terminology as an information asset
NIS2 Art. 20 (source) Governance (management body responsibilities) Approval workflows with mandatory review, documented processes, audit-ready evidence supporting management body approval and oversight of cybersecurity risk-management measures
NIS2 Art. 21 (source) Cybersecurity risk-management measures Approval workflow, audit trail, and documented terminology governance contribute to the policy and procedure evidence required under Art. 21(2)(a) and (g) (policies on risk analysis and information system security; basic cyber hygiene practices and training); they do not by themselves satisfy the full Art. 21 control set
FDA 21 CFR §11.10(e) (source) Audit trail (controls for closed systems) Append-only audit history table with user attribution and platform-supplied write timestamps for every create, edit, submit, approve, and reject action; record changes do not obscure previously recorded information; no delete/modify API is exposed
FDA 21 CFR §§11.50, 11.200 (source) Electronic signatures (manifestations; components and controls) Not implemented as a Part 11 electronic signature. The four-eyes approval workflow records submitter, reviewer, timestamp, and decision, but does not implement signature manifestations under §11.50 (printed name + signature meaning bound to the record) or the two-component identification, non-repudiation, and component-control requirements under §11.200. Customers operating under Part 11 must combine this audit trail with a separate Part 11-compliant electronic signature system
ALCOA+ Attributable Who did what, when Account ID + timestamp recorded on every create, edit, submit, approve, and reject action
ALCOA+ Contemporaneous Recorded at time of activity Real-time version tracking; audit entries created at the moment of each action, not retroactively
ALCOA+ Original First recording preserved Version 1 of every term is always retained; full version history is append-only at the app layer
ISO 13485 Cl. 4.2.4 Document control Four-eyes approval, versioning, status workflows (draft → review → approved → deprecated)

12. Vendor Information

Field Details
Legal entity DailyMind LTD
Registration number HE 439959
Jurisdiction Cyprus (European Union)
Security contact [email protected]
General contact [email protected]
Data Processing Agreement https://compliance-glossary.teamkit.dev/dpa
Privacy Policy https://compliance-glossary.teamkit.dev/privacy
Security Policy https://compliance-glossary.teamkit.dev/security
Marketplace listing Compliance Glossary for Confluence
Insurance No Professional Indemnity or Cyber Liability policy is in force today — disclosed up front so procurement can decide whether that fits their risk threshold. Certificate of insurance available on request once cover is bound.

Ready to Proceed?

This whitepaper covers the security and compliance posture of Compliance Glossary for Confluence. For additional questions, contact [email protected].

Evaluate in Confluence View Security Policy

Related Resources

Compliance for Confluence — approved terms, page scanning, and audit evidence for regulated teams in Confluence

Security Policy — infrastructure security, authentication, data protection, vulnerability reporting

Data Processing Agreement — GDPR-compliant DPA covering data processor obligations and sub-processors

Privacy Policy — what data we collect, how we process it, and your rights

About DailyMind LTD — the vendor, its registration, and contact routes

Product Limitations — what the app does not certify, approve, sign, or store

Compliance Guide — what auditors check and how we help, mapped to regulatory standards

Four-Eyes Principle — how the dual-approval workflow enforces segregation of duties