[REVIEW] container-security: add workload identity token audience scoping gates

[stmr]

Skill Being Reviewed

Skill name: container-security
Skill path: skills/cloud/container-security/

False Positive Analysis

Benign code that triggers a false positive:

Projected service account tokens set explicit audience and short TTL.

Why this is a false positive:
The current review guidance can push an agent to flag this pattern without first proving the attacker-controlled path, tenant boundary, or effective runtime behavior. The review should require evidence before classification.

Coverage Gaps

Missed variant 1:

Pod token audience is accepted by unintended cloud API or internal service.

Why it should be caught:
This is a realistic failure mode for container-security. It changes the effective security boundary without matching the simpler examples currently emphasized by the skill.

Missed variant 2:

Token gets mounted into sidecar/debug container that does not need it.

Why it should be caught:
This variant commonly appears in production systems and needs explicit reviewer prompts so agents do not stop at static configuration review.

Edge Cases

Kubernetes token defaults vary by version and cloud integration.

Remediation Quality

• Fix resolves the vulnerability
• Fix doesn't introduce new security issues
• Fix doesn't break functionality
• Issues found: Remediation guidance should add an evidence gate for the specific boundary, a benign exception pattern, and a regression check. Without those, fixes may be either cosmetic or over-broad.

Comparison to Other Tools

Tool	Catches this?	Notes
Semgrep	Partial	Can catch static patterns, but not effective policy, ownership, or runtime propagation without custom rules.
CodeQL	Partial	Strong for code/data-flow cases, weaker for cloud/control-plane and process evidence unless modeled.
Other: manual review	Yes	Human review can verify effective behavior, exception ownership, and operational evidence.

Overall Assessment

Strengths:
The skill gives useful practitioner framing and asks for concrete evidence instead of generic advice.

Needs improvement:
Need token audience/mount gates.

Priority recommendations:

1. Add a checklist item for workload identity token audience scoping evidence.
2. Add one benign exception example so reviewers avoid noisy findings.
3. Add one regression or verification step that proves the effective boundary after remediation.

Bounty Info

• I have read and agree to the CONTRIBUTING.md bounty terms
• Preferred payment method: PayPal samik4184@gmail.com