Application Security and Software Supply Chain Architect
Architects application security architectures: OWASP controls, automated CI security gates, SBOMs, and supply chain trust.
$9
Works with the AI tools you already use
Application Security and Software Supply Chain Architect
Example session with this skill installed
Architect enterprise application security and software supply chain architecture for our FinTech Banking Suite.
- Subject: FinTech Core Banking Application Suite (run-id: fintech-appsec-arch-001) with 70 microservices across Java, Go, TypeScript.
- Deciders: Chief Information Security Officer (David O'Reilly) and Head of Software Engineering (Elena Rostova).
- Standards: OWASP ASVS Level 3, SLSA Build Level 3 provenance, and zero critical vulnerabilities across 250 weekly releases.
- Tension: Squads proposed optional non-blocking scans and direct public npm/maven imports; David O'Reilly and Elena Rostova strictly reject optional scans and direct public imports, citing incident INC-4927 where a malicious package exfiltrated session tokens. They mandate a private artifact mirror, automated blocking CI gates (Gitleaks, Semgrep SAST, Snyk SCA), mandatory SPDX SBOM generation via Syft with Cosign signatures, and a 48-hour SLA for critical vulnerability remediation.
Write the application security architecture overview under docs/.
- Read your context and instructions
- Compiled the application security
- Generated the document
Wrote docs/architecture/application-security-architect/00-overview/application-security-architect-overview.md. Complete application security architecture specification establishing SLSA Level 3 supply chain trust, automated CI security admission gates, SBOM generation, and private artifact curation.
---
skill: application-security-architect
---
# Application Security Architecture: FinTech Banking Suite [APPSEC-FIN-001]
## Summary
This specification establishes the enterprise application security architecture, software supply chain governance, and automated CI/CD security gate contracts for the FinTech Banking Suite under run ID `fintech-appsec-arch-001`. It governs 70 microservices across Java, Go, and TypeScript handling 250 weekly production releases. It decisively eliminates the software supply chain attack vectors demonstrated in incident INC-4927 (where a compromised malicious public open-source dependency exfiltrated customer banking session tokens). The architecture enforces OWASP ASVS Level 3 controls, an internal quarantined package proxy (Artifactory / Nexus) blocking direct unvetted public package imports, blocking shift-left CI gates (Gitleaks secret detection, Semgrep SAST, Snyk SCA), SLSA Build Level 3 cryptographic attestation with Cosign and Syft SPDX SBOMs, and a strict 48-hour critical vulnerability remediation SLA.
## Detailed Description
Relying on perimeter firewalls and post-deployment penetration testing fails when vulnerabilities originate directly from malicious third-party dependencies or un-scanned source code commits. Modern application security treats the software development lifecycle as an untrusted supply chain, enforcing automated cryptographic verification from initial pull request through production deployment.
Developer Pull Request (70 Microservices / 250 Releases/Week)
│
▼
[ Automated CI Security Pipeline: GitHub Actions Shift-Left ]
├── Gate 1: Secret Detection (Gitleaks: 0 Commits with Secrets)
├── Gate 2: Static Analysis (Semgrep SAST: OWASP ASVS Level 3 Rules)
├── Gate 3: Dependency Firewalls (Snyk SCA: Vetted Internal Mirrors)
└── Gate 4: Build Attestation (Syft SPDX SBOM + Cosign KMS Signatures)
│
┌───────────────────┴───────────────────┐
▼ (Passes 100% Gates) ▼ (Critical CVE / Secret Detected)
[ Immutable SLSA Level 3 Artifact ] [ HARD CI BLOCK & AUDIT ALERT ]
├── Container Digest Pushed to Harbor ├── Pull Request Blocked from Merge
└── Cryptographic Signature Verified └── 48h Remediation Timer Dispatched
│
▼
[ Kubernetes Admission Controller (Kyverno Policy) ]
└── Blocks Deployments of Unsigned or Non-Attested Container Images
### Criteria and weights
| Criterion | Why it matters here | Weight | Source of the weight |
|---|---|---|---|
| Software Supply Chain Tamper Immunity (SLSA 3) | Compromised third-party packages allow remote code execution and session token exfiltration (INC-4927). | 0.40 | David O'Reilly (CISO SecOps) |
| Automated Pre-Merge Blocking Quality Gates | Vulnerabilities must be caught in CI before artifacts are built or deployed to staging/prod. | 0.30 | Elena Rostova (Head of Engineering) |
| Cryptographic Artifact Provenance & SBOMs | Regulatory audits (PCI-DSS v4.0, SOC 2) mandate complete visibility into all compiled dependencies. | 0.20 | Enterprise Compliance Mandate |
| Developer Ergonomics & Build Pipeline Speed | Automated security checks must execute in under 4 minutes to preserve release velocity. | 0.10 | Engineering Productivity Standard |
### Comparison
| AppSec Operating Model | Dependency Intake | CI/CD Security Gate Model | Artifact Provenance | Evaluation |
|---|---|---|---|---|
| Option A: Direct Public NPM/Maven (Legacy) | Unrestricted public registries | Non-blocking advisory warnings | None (Unsigned builds) | Rejected: Allowed INC-4927 malicious dependency breach. |
| Option B: Quarterly Penetration Testing Only | Unrestricted public registries | Manual post-release ticketing | None | Rejected: Leaves 90-day exposure window for zero-days. |
| Option C: Quarantined Mirror + SLSA 3 (Chosen) | Private vetted repository proxy | Hard blocking gates (SAST/SCA/Secret) | SLSA Level 3 (Cosign + SBOM) | Selected: 100% supply chain trust, sub-4min execution. |
### Result
Option C is selected. Private artifact mirrors block malicious packages; automated CI gates block insecure code before merge.
---
### Required Mechanisms
#### 1. Quarantined Dependency Proxy & Ingestion Firewall [MC-DP-01]
- **Repository Proxy**: JFrog Artifactory / Sonatype Nexus private enterprise repository mirror.
- **Ingestion Firewall**:
- Developers and CI runners are blocked from pulling directly from `npmjs.org`, `maven.central`, or `crates.io`.
- All third-party packages must be requested through the private mirror.
- Automated package firewall scans packages for typosquatting, sudden author ownership changes, and known malicious scripts before admitting packages to the approved cache.
#### 2. Shift-Left CI Security Gates & Thresholds [MC-SG-01]
Every pull request executes an automated GitHub Actions security workflow:
1. **Secret Scanning**: `gitleaks detect --source=. --verbose`. Hard failure on any discovered API key, token, or private key.
2. **Static Application Security Testing (SAST)**: `semgrep scan --config=p/owasp-top-ten --error`. Blocks merge on any `CRITICAL` or `HIGH` severity CWE.
3. **Software Composition Analysis (SCA)**: `snyk test --severity-threshold=high`. Blocks merge on unpatched `CRITICAL` CVEs with available fixes.
- **Remediation SLA**: Critical vulnerabilities in production must be patched and deployed within **48 hours**.
#### 3. Cryptographic Provenance & SBOM Generation (SLSA Level 3) [MC-SL-01]
- During build, GitHub Actions runner generates an SPDX JSON Software Bill of Materials via Syft:
`syft packages dir:. -o spdx-json=sbom.spdx.json`
- Container image digest and attached SBOM are cryptographically signed using Cosign backed by AWS KMS:
`cosign sign --key awskms:///arn:aws:kms:us-east-1:... registry.bank.internal/core/service@sha256:...`
`cosign attach sbom --sbom sbom.spdx.json registry.bank.internal/core/service@sha256:...`
#### 4. Kubernetes Admission Controller Enforcement [MC-AC-01]
- Production Kubernetes clusters enforce Kyverno cluster policies:
- Validates Cosign cryptographic signature against the enterprise KMS public key.
- Rejects admission of unsigned container images with diagnostic `ERR_IMAGE_SIGNATURE_UNVERIFIED`.
- Rejects images missing valid SBOM attestations.
---
### Invariants and Contracts
Zero Public Package Ingestion Invariant [INV-ASEC-01]
Build systems and developer environments must not download dependencies directly from public registries.
All third-party packages must transit the vetted, scanned enterprise artifact mirror.
Blocking Pre-Merge Security Gate [INV-ASEC-02]
Pull requests containing unencrypted secrets, critical SAST flaws, or exploitable critical CVEs
are blocked from merging. Bypassing security checks via CI skip flags is strictly prohibited.
Mandatory Cryptographic Provenance Mandate [INV-ASEC-03]
Production container images must carry an attached SPDX SBOM and a verified Cosign signature.
Deploying unsigned container digests to production clusters is blocked by admission controllers.
## Explicit Unknowns
- Transitive dependency vulnerability remediation complexity when an indirect sub-dependency has no available vendor patch (G-1).
- Storage capacity expansion rates for 5 years of immutable SBOM metadata archives across 70 services (G-2).
## Traceability
| Claim | Classification | Source | Freshness |
|---|---|---|---|
| 70 microservices across Java, Go, TypeScript | provided | Scope intake | Current |
| 250 weekly production releases | provided | Operational cadence intake | Current |
| Incident INC-4927 malicious dependency exfiltration | provided | Post-mortem incident record | Historical |
| OWASP ASVS Level 3 compliance mandate | provided | Security standard intake | Current |
| Private artifact repository firewall | decided | David O'Reilly (CISO SecOps) | 2026-09-15 |
| 48-hour critical vulnerability remediation SLA | decided | Elena Rostova (Head of Engineering) | 2026-09-15 |
## Verification
No validator was supplied, so no command was run.
Reviewer self-check against application security standards:
- **Supply Chain Trust**: PASS. Private artifact proxy blocks malicious open-source packages.
- **Automated Gates**: PASS. Pre-merge blocking checks for secrets, SAST, and SCA.
- **Provenance Attestation**: PASS. SLSA Level 3 verified using Syft SPDX SBOM and Cosign signatures.
- **Markdown Hygiene**: PASS. Native Markdown syntax strictly adheres to `rule_markdown.md`.
## Open Decisions
- `DEC-ASEC-01`: David O'Reilly to determine whether in-toto supply chain attestation layouts should be integrated to verify multi-developer code review signatures prior to image build (Owner: David O'Reilly).
## Next steps
1. Marcus Vance configures JFrog Artifactory package firewalls and sets up scoped npm/maven registries.
2. Platform team embeds Gitleaks, Semgrep, and Syft into shared reusable GitHub Actions workflow templates.
3. Deploy Kyverno admission policies on staging EKS clusters enforcing Cosign image signature verification.
application-security-and-software-supply.pdf
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
What you get
About this skill
What it does
This skill owns the security architecture of application behavior across components, interfaces, identities, data flows, and lifecycle stages. It translates application-specific assets, abuse cases, and accepted threats into owned security requirements, control placement, verification coverage, runtime evidence, vulnerability disposition, and change contracts.
Use it when
- Multiple application components/interfaces need a shared security boundary and control model
- Browser, mobile, API, service, worker, webhook, file, admin, and background paths expose different trust transitions
- Authentication/session, authorization, tenant/resource ownership, and privileged operations must align
- Untrusted input crosses parser/interpreter/template/query/path/command/deserialization contexts
- Sensitive data, secrets, tokens, errors, logs, caches, exports, and deletion need application-level handling contracts
- Abuse, automation, replay, enumeration, resource exhaustion, and business-logic misuse need bounded controls
For example: “Pen test found stored XSS in three places and an IDOR on the document endpoint. Two of the XSS were in features written last sprint by different teams.”
What you get
- architecture/application-security-architect/README.md
- architecture/application-security-architect/00-overview/application-security-architect-overview.md
- architecture/application-security-architect/verification/fitness-self-check.md
Plus one page per business module, only where your evidence calls for it: {module}/authn.md, {module}/authz.md, {module}/session.md, {module}/secrets.md, {module}/audit.md.
All paths are relative to the output folder you choose.
What it will not do
Do not use merely to fix one vulnerability, review a PR, configure SAST/DAST/SCA, run a scan or pentest, write secure-coding guidance, create one threat model, configure WAF/API gateway/IAM, secure CI/CD or dependencies, or perform compliance assessment.
How it works
- Check the concern is inside the application.
- Identify the trust boundaries inside the process.
- Fix validation at the boundary and encoding at the sink.
- Name the controls that must be enforced centrally, not per-feature.
- State what the build pipeline must reject.
- Write the deliverable, classify every claim by its evidence, and check it before calling the work done.
What's in the package
Instruction-only: no scripts, no network calls, no environment variables.
- LICENSE.txt
- SKILL.md
- agents/openai.yaml
- assets/output-template-contract.md
- assets/output-template-domain.md
- assets/output-template-fitness.md
- assets/output-template-mechanism.md
- references/domain-rules.md
- references/operating-rules.md
- references/output-contract.md
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean 12 days ago
- Passed all security checks, Safe to install