Skip to content

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security

chore(codex-security): sync public projection

ianw-oai132940443c1e
diff --git a/sdk/typescript/_bundled_plugin/references/static-finding-assessment.md b/sdk/typescript/_bundled_plugin/references/static-finding-assessment.md
index dbe2a23a9..7f1d11f83 100644
--- a/sdk/typescript/_bundled_plugin/references/static-finding-assessment.md
+++ b/sdk/typescript/_bundled_plugin/references/static-finding-assessment.md
@@ -47,10 +47,7 @@ Before treating a static path as security-relevant, classify:
- policy basis: repository policy, product docs, deploy/config evidence,
package metadata, code comments, threat model, or unknown
-A reachable dataflow is not enough on its own. Static evidence is strongest when the vulnerable condition is reachable and the source crosses a product security boundary the project appears to support.
-
-If a path is only example-only, fixture-only, docs-only, generated-only,
-vendored-only, local-only, trusted configuration, or intentionally code-executing extension behavior, record that as counterevidence unless other repository evidence shows the surface is shipped, deployed, documented for untrusted users, or bypasses a supported hardening/auth boundary.
+Check whether untrusted input can reach the bug.
## Static Confidence
diff --git a/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/SKILL.md b/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/SKILL.md
index a40bf9a95..3d5adbfcb 100644
--- a/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/SKILL.md
+++ b/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/SKILL.md
@@ -20,7 +20,7 @@ Use the shared scan artifact path conventions in `../../references/scan-artifact
1. Load the per-scan threat model path from `../../references/scan-artifacts.md` as the repo-specific threat-model source of truth. Start from this along with the potential findings. Both inputs are required for this workflow.
- For repository-wide and scoped-path scans, include validation closure rows marked `reportable` or `survives: yes` even if they were not assigned polished candidate numbers during discovery.
-2. Determine whether the affected code is in scope for the repository threat model and whether it belongs to a real product surface or real production workflow.
+2. Determine whether the affected code is in scope for the repository threat model and whether it belongs to a product surface or production workflow.
3. Build a factual attack path using repository evidence only:
- service mapping
- exposure and entry points
@@ -40,7 +40,7 @@ Use the shared scan artifact path conventions in `../../references/scan-artifact
Use this checklist before finalizing the attack-path facts or policy decision:
- Determine whether the finding is actually a real security vulnerability rather than a correctness bug or false positive.
-- Determine whether the affected code belongs to a real product surface or meaningful production workflow.
+- Determine whether the affected code belongs to a product surface or production workflow.
- Map the relevant service, component, or workflow context from repository evidence.
- Establish exposure and entry points from repository evidence such as listeners, ingress, load balancers, service ports, manifests, routing, or network policy.
- Establish identities, privileges, and trust boundaries that matter for the path.
@@ -103,5 +103,5 @@ Render attack-path facts using `references/attack-path-facts.md`.
- Save a final visible report for each candidate finding using that finding's attack-path analysis report path from `../../references/scan-artifacts.md`.
-- Considerations for attack path --
-- A finding should count as a real security issue if a realistic attacker could use it from a reasonable attack surface relevant to the product, especially if it is something that is part of the thread model.
-- The attack surface should generally be one that is plausibly exposed to end users / external actors (or another actor explicitly in scope in the threat model), not an arbitrary internal-only or contrived path.
+- A bug matters if evidence shows an attacker could exploit it.
+- The attack surface should generally be one that is plausibly exposed to end users / external actors (or another actor explicitly in scope in the threat model).
diff --git a/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/references/severity-policy.md b/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/references/severity-policy.md
index f5a96d492..d3f31f03b 100644
--- a/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/references/severity-policy.md
+++ b/sdk/typescript/_bundled_plugin/skills/attack-path-analysis/references/severity-policy.md
@@ -53,7 +53,7 @@ Examples that usually should not remain `high`/`critical` without very strong pr
- Generic correctness/reliability bugs
- Strange edge cases with unclear attacker value
- Low-impact information leaks
-- Internal-only defects without realistic attacker reachability
+- Internal-only defects without attacker reachability
- The report shows a bug class in isolation, but not a realistic exploit path.
- The issue requires the attacker to already have privileged, admin, root, console, shell, or code-execution access.
- "Could maybe matter if chained with many assumptions" arguments
@@ -105,7 +105,6 @@ Final policy-adjustment guidance:
- `localhost` usually supports low likelihood unless a lower-privileged attacker can realistically reach that listener
- `none` does not increase likelihood based on exposure
- Then decide reportability using the existing facts:
- - if the component is not part of a real product surface or meaningful production workflow, set `ignore`
- if repository evidence does not establish a realistic lower-privileged in-scope attacker path, set `ignore`
- if the path is internal-only, developer-only, operator-only, localhost-only, privileged-local, or otherwise not meaningfully reportable in context, set `ignore`