DNA by Humaan is an open-source project intelligence platform: CLI tooling, a runtime observer, GitHub automation, a knowledge-pack marketplace, and AI-assisted repair for TypeScript teams. We take security seriously and welcome responsible disclosure from researchers, customers, and the community.

This policy covers the DNA monorepo (superhumaan/DNA), published npm packages, and closely related first-party surfaces. It does not replace your own organisation's security programme for projects scaffolded or managed by DNA.

• Supported versions
• Scope and architecture
• Reporting a vulnerability
• In scope
• Out of scope
• Research guidelines
• Severity and response timelines
• Disclosure and safe harbour
• Recognition
• How we secure DNA
• Supply chain and package integrity
• Credential and secret handling
• Data handling and privacy
• Security guidance for DNA users
• Security updates
• Related repositories
• Questions

Security fixes are applied to the latest release on npm. Development builds on main may receive fixes before the next tagged release; we recommend staying on published releases for production use.

Check your installed version:

dna --version
# or
npm ls @superhumaan/dna-by-humaan

DNA spans local developer tooling, optional cloud-adjacent services, and generated project artifacts. The table below maps components to their security relevance.

DNA maintains two worlds in managed projects:

• .DNA/ — machine-readable intelligence (behaviour, memory, runtime events, plans)
• DNA/Impressions/ — human-facing documentation (product, architecture, security, compliance)

Do not open a public GitHub issue for security vulnerabilities. Public disclosure before we have had a chance to investigate puts users at risk.

1. GitHub private vulnerability reporting (preferred)
   Report a vulnerability on this repository's Security tab.

2. Email
   Send details to security@humaan.com with the subject line DNA Security Report.

For issues specific to the hosted web application or marketplace API, you may also report via the same channels and note DNA-Web in the subject or body. We coordinate across repositories when needed.

The more detail you provide, the faster we can triage and reproduce.

Subject: DNA Security Report — [short title]

Summary:
[What is the vulnerability?]

Component:
[e.g. dna-github OAuth flow / runtime redaction / marketplace install]

Impact:
[What can an attacker do?]

Steps to reproduce:
1.
2.
3.

Environment:
- DNA version:
- Node.js version:
- OS:

Proof of concept:
[Attach or describe]

Credit:
[Your name/handle, or "anonymous"]

1. Acknowledgement — we confirm receipt.
2. Triage — we reproduce, assess severity, and identify affected versions.
3. Remediation — we develop and test a fix on a private branch.
4. Release — we publish a patched npm release and GitHub Security Advisory.
5. Disclosure — coordinated public disclosure after users can update (see Disclosure and safe harbour).

We will keep you informed of progress. If you prefer to remain anonymous, say so in your report.

The following classes of vulnerability are in scope when they affect DNA tooling, published packages, first-party services, or unsafe DNA defaults in generated artifacts.

• Remote code execution (RCE) in the CLI, runtime, or install/generation pipeline
• Arbitrary command execution via malicious marketplace packs, templates, or IVF output without explicit user action
• Sandbox escape or privilege escalation beyond the invoking user's OS permissions
• Authentication bypass in DNA-managed GitHub or web flows

• Leakage of GITHUB_TOKEN, OAuth tokens, or credentials stored in ~/.config/dna/
• Writing secrets to .DNA/runtime/ events, logs, GitHub issues, or AI repair payloads without redaction
• Insecure transmission of tokens (e.g. logged in CI output, embedded in generated workflow files by default)
• Token scope escalation beyond declared repo, workflow, read:org usage

• Flaws in OAuth device flow, token refresh, or credential file permissions (expected: directory 0700, file 0600)
• Unsafe git push URL construction exposing tokens in remotes, logs, or error messages
• Unauthorised repository access, issue creation, or workflow modification

• Failure to redact sensitive patterns (Bearer tokens, API keys, passwords, cookies) before persistence
• Path traversal or arbitrary file write via .DNA/runtime/ or runtime DB storage
• Exfiltration of stack traces or request bodies containing PII beyond what the user configured

• Integrity failures in bundled or remote catalog metadata (tampered pack contents, signature bypass)
• Install path traversal allowing writes outside .DNA/knowledge/
• Malicious default content in bundled STEM, compliance, or platform packs distributed by Humaan

• Unsafe defaults in generated CI/CD, Docker, git hooks, or feature-factory templates that expose secrets or disable security controls without opt-in
• dna init, dna ivf, or dna plan output that embeds credentials, private keys, or hardcoded tokens
• Doctor/validate false negatives that silently approve known-dangerous configurations

• Prompt injection leading to execution of attacker-controlled shell commands via DNA repair workflows
• Unauthorised disclosure of repository contents, .DNA/ memory, or private GitHub data to external AI providers beyond documented behaviour

• Vulnerabilities in dna.humaan.app or the marketplace API hosted from superhumaan/DNA-Web
  Report via the same channels; we route to the appropriate maintainers.

The following are generally out of scope unless they demonstrate a flaw in DNA itself (not merely in a consumer project or upstream dependency):

If you are unsure whether something is in scope, report it anyway. We would rather receive a borderline report than miss a real issue.

We support good-faith security research. Please follow these rules:

• Test against your own installations, forks, or local environments.
• Use minimal proof-of-concept code to demonstrate impact.
• Allow reasonable time for remediation before public disclosure.
• Redact third-party secrets, PII, and customer data from reports.
• Stop testing when you have confirmed the vulnerability.

• Access, modify, or delete data belonging to other users or Humaan production systems without authorisation.
• Perform destructive testing (data wipe, service disruption, resource exhaustion) against shared infrastructure.
• Spam reports, use automated scanners without manual validation, or submit duplicate reports for known issues.
• Exploit a vulnerability beyond what is necessary to demonstrate it.
• Publicly disclose before we have published a fix or agreed on a disclosure date.

We assess severity using CVSS v3.1 where applicable, combined with exploitability and blast radius in the DNA ecosystem.

Timelines are targets, not guarantees. Critical issues are prioritised above all other work. We will communicate delays and provide interim mitigations when a full fix takes longer.

We ask reporters to:

1. Report privately through the channels above.
2. Allow us 90 days from acknowledgement to release a fix (we may agree on a shorter or longer window depending on severity and complexity).
3. Avoid public disclosure, blog posts, or social media posts until a fix is available — unless we agree otherwise or disclosure is required by law.

When a fix is ready, we publish a GitHub Security Advisory with credit to the reporter (unless you request anonymity) and update CHANGELOG.md.

We support responsible disclosure. We will not pursue legal action against researchers who:

• Make a good-faith effort to avoid privacy violations, data destruction, and service disruption.
• Give us reasonable time to investigate and remediate before public disclosure.
• Do not exploit the issue beyond what is needed to demonstrate it.
• Comply with the Research guidelines above.

This safe harbour applies to research conducted in accordance with this policy. It does not authorise access to systems or data you do not own or have permission to test.

We are grateful to security researchers who help keep DNA safe. With your permission, we will:

• Credit you in the GitHub Security Advisory.
• Mention your handle in the CHANGELOG security section for significant findings.

We do not currently operate a paid bug bounty programme. If that changes, we will update this policy.

We follow security practices appropriate for open-source developer tooling and Humaan's broader secure SDLC.

Always install DNA from official sources:

npm install @superhumaan/dna-by-humaan
# or
pnpm add @superhumaan/dna-by-humaan

Confirm the package scope and publisher on npmjs.com. The canonical source repository is github.com/superhumaan/DNA.

Knowledge packs install into .DNA/knowledge/ from:

1. The hosted catalog at dna.humaan.app/marketplace
2. A bundled offline catalog in @superhumaan/dna-by-humaan when remote is unavailable

Review pack contents after install, especially in regulated environments. Report supply-chain or integrity issues through the reporting channels above.

In scope: tampered npm tarballs, compromised publish credentials, malicious bundled catalog entries, or MITM attacks against the marketplace API that alter pack content.

@superhumaan/dna-by-humaan is a developer CLI and runtime observer, not a passive config library. Security scanners (including Socket.dev) may flag the following behaviours — all are documented and intentional:

Not shipped: install scripts (preinstall / postinstall), telemetry packages, recursive self-dependencies, or third-party production npm dependencies on @superhumaan/dna-by-humaan (v0.4.5+). The published tarball bundles internal implementations for CLI parsing, config validation, file globbing, git, GitHub REST, logging, and file watching. Optional peer dependencies (express, fastify, @nestjs/common) are declared for runtime middleware adapters only — your app already installs the framework you use.

Offline fallbacks: bundled assets/marketplace-catalog.json and assets/intelligence-catalog.json when remote catalogs are unreachable.

Published releases use npm provenance from GitHub Actions. Verify integrity:

npm view @superhumaan/dna-by-humaan --json | jq '.dist.integrity'

Report unexpected network destinations or install-time behaviour through the reporting channels above.

Socket may flag optional framework peer dependencies (express, fastify, @nestjs/common) when reviewing DNA's package metadata. These peers apply only when you wire runtime middleware; DNA does not install them automatically.

A common example is obfuscated code in next — Vercel ships pre-compiled, minified React Server Components bundles. Static analyzers classify this as obfuscated; it is normal production build output, not a DNA supply-chain compromise.

DNA's Next.js runtime adapter uses duck-typed request interfaces and does not import next at install time. Next.js projects already declare next in their own package.json; DNA generates middleware snippets that run in the consumer app.

DNA interacts with credentials in several places. Researchers and users should understand the expected behaviour.

If you discover credentials written to an unexpected location, or redaction bypassed, that is in scope.

DNA is designed to run primarily on developer machines and in consumer project repos. Runtime events, cellular memory, and knowledge packs stay in .DNA/ unless you explicitly enable integrations (e.g. GitHub issue creation, AI repair).

The runtime observer applies pattern-based redaction before persisting messages and stack traces:

• Bearer tokens
• API keys
• Passwords
• Generic token= / authorization= / cookie= values

Redaction is not a guarantee against all PII or custom secret formats. Report gaps where sensitive data persists in .DNA/runtime/ or exported GitHub issues.

DNA ships scrubbed GDPR and compliance document templates with [Company Name] / [Product Name] placeholders. These are templates for your organisation — not Humaan processing your data. See compliance tiers for how DNA coordinates compliance implementation.

When AI repair is enabled, project context may be sent to configured AI providers. Understand your provider's data handling before enabling this in production or with sensitive codebases.

These steps reduce risk in projects managed by DNA:

# Verify installation health
dna doctor
dna validate

# Check for quality/security patterns
dna quality report

# Keep DNA updated
npm update @superhumaan/dna-by-humaan
# or
dna update

Recommendations:

• Pin DNA to semver ranges and review CHANGELOG before major upgrades.
• Never commit .env, github-credentials.json, or tokens in .DNA/.
• Use fine-grained GitHub tokens or OIDC in CI instead of long-lived PATs where possible.
• Review generated CI, Docker, and git-hook files before merging — especially in regulated environments.
• Keep github.enabled and aiRepair.enabled off in runtime until you understand data flows.
• Run dna marketplace install only for packs you trust; audit .DNA/knowledge/ in sensitive repos.
• Add dna doctor and dna validate to your CI pipeline (see CI/CD docs).

When we fix a vulnerability:

1. A GitHub Security Advisory is published with severity, affected versions, and remediation steps.
2. A patched release is published to npm under @superhumaan/dna-by-humaan (and sibling packages as needed).
3. The fix is noted in CHANGELOG.md.

Subscribe to advisories: watch this repository's Security Advisories or enable GitHub security alerts for your projects that depend on DNA.

After an advisory: upgrade to the latest npm release and re-run dna doctor in affected projects.

Last updated: July 2026