A.8.4 Technological
Access to source code
Structured attributes from ISO/IEC 27002:2022 — control type · CIA properties · cybersecurity concept · operational capability · security domain. What do these mean?
Mapped NIST 800-53 r5 controls (17)
Our reading (mostly llm_verified) of how this ISO control and each NIST 800-53 control relate. Not an ISO or NIST product.
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- CM-5partialcovers — A.8.4's narrow focus on source-code access restrictions addresses only a slice of CM-5's broader requirement to restrict all physical/logical changes to the entire system
Aligned NIST CSF 2.0 outcomes (18)
NIST CSF 2.0 outcomes this ISO control aligns with — our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- PR.AA-05domain mandateimplements — A.8.4's technical requirement to control (restrict) access to source code directly operationalizes the management and enforcement of access permissions within PR.AA-05's governance outcome, reaching it through the shared access-control domain even though the CSF outcome does not name source code specifically
Related OWASP ASVS 5.0 requirements (5)
Application-security verification requirements (OWASP ASVS 5.0) this ISO control aligns with; links open the ASVS chapter. Our analysis (mostly llm_verified) — many ISO controls have no ASVS counterpart.
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Related weaknesses / CWE (3239)
Weakness classes this ISO control helps prevent or mitigate — our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- CWE-508partialprevents — A.8.4's stated purpose explicitly includes preventing introduction of unauthorized functionality and malicious changes via source code, and its write-access restrictions, change-control linkage, and audit logging address the Implementation-phase introduction path; however, malicious code can also arrive through authorized insiders, third-party/open-source components, compiled binaries, and runtime vectors outside the source repository, so a nameable remainder of introduction paths is untouched.
- CWE-511partialblocks — Restricting access to source code limits opportunities for insiders to plant logic/time bombs.
- CWE-511partialprevents — A.8.4 restricts who may write source code, so logic bombs that would have been planted by unauthorized personnel are kept out; however, authorized developers can still introduce hidden malicious logic, and the clause does not review code content, leaving the insider and compromised-toolchain paths open.
- CWE-540partialblocks — A.8.4 restricts who can read source code in repositories and via development tools, which blocks exploitation through direct repository access; however, CWE-540's observed examples are predominantly about source code being served through applications (client-side JavaScript tokens, JSP comments in rendered HTML, .inc files served without preprocessing), which A.8.4's repository-access mechanism does not reach.
- CWE-862partialblocks — Mandatory authorization checks before allowing modifications ensure that every sensitive action on source code is preceded by an explicit access-control decision.
Mitigated MITRE ATT&CK techniques (1865)
Adversary techniques (MITRE ATT&CK Enterprise) this ISO control helps mitigate; links open attack.mitre.org. Our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- T1195mostlyprevents — Restricting write access to source code repositories and requiring change-control authorization reduces the likelihood that an adversary can insert malicious code into the organization's software supply chain.
- T1213.003mostlyprevents — A.8.4 directly controls the mechanism T1213.003 exploits — read/write access to source code repositories, audit logging, and role-based restrictions prevent unauthorized collection of source code and embedded credentials; the named remainder is access via compromised privileged accounts or third-party-hosted repositories outside the clause's scope.
- T1080partialdetects — A.8.4's audit log requirement (item f) records all accesses and changes to source code, which provides a forensic trail that could surface tainting of an internal code repository; however, the clause's scope is source code repositories and development tools, not network drives or general shared storage, and an audit log records the change without flagging it as malicious — detection requires review and is limited to the source-code portion of T1080's surface.
- T1080partialprevents — A.8.4's write-access restrictions, change-control procedures and audit logging directly prevent adversaries from tainting source code in repositories the control governs, but T1080 also covers network drives, file shares and other shared storage that fall outside A.8.4's scope, so only a slice of the technique's surface is prevented.
- T1127.003partialprevents — A.8.4 restricts write access to source code repositories and controls access to development tools (which includes build utilities like JamPlus), making it harder to introduce a malicious .jam file into a build pipeline or for unauthorized users to invoke the tool; however, it does not prevent JamPlus from being installed or used by authorized developers (or under a compromised developer account) directly on a workstation, so the technique remains reachable through that path.
- T1195.001partialprevents — A.8.4 restricts write access to source code and mandates change control, which prevents an adversary from directly injecting malicious code into the organization's own repositories, but it does not prevent the upstream compromise of an external dependency or development tool before it is ingested.
- T1195.002partialprevents — A.8.4 restricts and audits write access to source code, which prevents the insider-source-code-manipulation vector of T1195.002, but the technique is broader: it also covers manipulation of update/distribution mechanisms, replacement of compiled releases, and compromise at third-party suppliers and build pipelines — none of which A.8.4 reaches, since the technique explicitly occurs prior to receipt by the final consumer.
- T1195.002partialdetects — A.8.4's audit log of source code accesses and changes (clause f) detects manipulation of source code within the repository, which is one of the three vectors T1195.002 enumerates; it does not detect manipulation of update/distribution mechanisms or replacement of compiled releases, and a sophisticated actor with valid write credentials can still make changes that look legitimate in the log.
- T1213partialprevents — A.8.4 directly controls read/write access to source code repositories (a named sub-target of T1213, which explicitly lists 'source code snippets' and 'code repositories' among its examples), so unauthorized mining of source code via repositories is prevented where the control is in force; however, T1213 spans a much broader surface — SharePoint, Confluence, databases, CRM, messaging platforms, and many non-code data types (policies, diagrams, credentials, schedules, PII) — that A.8.4 does not address, leaving the bulk of the technique reachable.
- T1213.002partialprevents — A.8.4 restricts write access to source code and central repositories, which prevents malicious modification of code stored in SharePoint, but it does not prevent an adversary with read access from mining the information already present in the repository.
- T1213.003partialdetects — A.8.4's clause (f) requires an audit log of all accesses and changes to source code, which provides some detection capability for unauthorized repository access underpinning T1213.003, but the control's primary mechanism is preventive access restriction and the log records access events rather than the broader collection activity, leaving a bounded remainder.
- T1552.001partialprevents — A.8.4 restricts who can read source code and mandates audit logs and change control, which reduces the population able to search source code for embedded credentials and constrains unauthorized additions of credential-bearing code; however, it does not prevent developers from embedding credentials in source code in the first place, and the technique extends well beyond source code to configuration files, user credential stores, and cloud/container artifacts that A.8.4 does not address.
- T1565partialprevents — A.8.4 prevents unauthorized write access to source code and associated items, which blocks the subset of T1565 where the manipulated data IS source code (e.g., inserting malicious logic into a codebase), but T1565 covers manipulation of business data, database records, configuration, and other non-source-code artifacts that fall outside A.8.4's scope.
- T1565.001partialprevents — A.8.4 controls read/write access to source code repositories and associated development items, which prevents unauthorized stored-data manipulation of source code specifically; it does not address the broader T1565.001 surface (databases, emails, Office files, custom formats), and even within source code, privileged writers retain the ability to manipulate.
- T1589.001partialprevents — A.8.4 restricts access to source code repositories and associated items, which prevents credential gathering via that specific vector (e.g., hardcoded secrets in code, commit history exposure), but the technique spans many vectors — phishing, breach dumps, dark-web markets, infostealer logs, MFA-service compromise — that A.8.4 does not reach.
- T1677partialprevents — A.8.4's write-access restrictions and change-control procedures prevent Direct and Indirect Pipeline Execution by blocking unauthorized commits to the repository, but the Public Pipeline Execution scenario does not require write access at all — the adversary triggers the CI pipeline from a forked pull request, which A.8.4's source-code access controls do not reach, and privileged insiders with legitimate write access remain a vector.
- T1698partialprevents — A.8.4 restricts write access to source code repositories, blocking one of T1698's listed delivery vectors (code), but the technique has many other vectors (web pages, documents, emails, tickets) the control does not touch, and even within the code vector the control does not stop an LLM from processing already-injected prompts in code it can read.
Prevented OWASP Web Top 10 (2025) risks (2)
OWASP Web Top 10 (2025) risk categories this ISO control helps prevent or mitigate — our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- A01partialblocks — A.8.4 strictly controls read/write access to source code and development tools per roles and change procedures, which prevents many Broken Access Control defects (e.g. missing function-level checks, IDOR via improper authorization in code) from being introduced during development; it does not address runtime enforcement failures, path traversal outside source, or CSRF.
- A08partialblocks — A.8.4's strict access controls on source code and libraries (read/write separation, change control, no direct repo access, audit logs) directly block unauthorized or malicious modifications that introduce integrity failures, covering several members of A08 such as unsigned updates and CI/CD compromise paths, but not all (e.g. insecure deserialization of untrusted data).
Control IDs, short titles and the structured attribute table (control type, CIA properties, cybersecurity-concept, operational capability, security domain) are facts from ISO/IEC 27001:2022 Annex A / ISO/IEC 27002:2022. The full implementation guidance prose lives in ISO/IEC 27002:2022 — not reproduced here. Cross-walks to NIST 800-53, NIST CSF 2.0, OWASP ASVS, CWE, MITRE ATT&CK and OWASP Web Top 10 are our own analysis (mostly llm_verified), not an ISO, NIST, MITRE or OWASP product — how ours compare.