Most external-command alerts are Markdown code fences, not executable backtick expressions. The audit confirms the remote Linkerd installer piped into sh as a critical supply-chain risk. It also flags an overbroad unauthenticated authorization example that should be narrowed.
curl --proto '=https' --tlsv1.2 -sSfL https://run.linkerd.io/install | sh
The install template pipes a remote script from https://run.linkerd.io/install directly into sh. This executes network-delivered code without checksum or review, creating supply-chain and command-execution risk.
The ServerAuthorization template allows unauthenticated clients from 10.0.0.0/8. This can teach overly broad ingress trust if copied without narrowing networks or using meshTLS identities.
The unauthenticated setting and broad private CIDR are explicit in the policy example. The text frames it as ingress traffic, so legitimate use is possible but it needs stronger qualification.
Capability review items (1)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
curl --proto '=https' --tlsv1.2 -sSfL https://run.linkerd.io/install | sh
The URL retrieves the Linkerd installer during the installation workflow. It is a legitimate project URL, but it still introduces network-sourced code into the setup path.
Most external-command alerts are Markdown code fences, not executable backtick expressions. The audit confirms the remote Linkerd installer piped into sh as a critical supply-chain risk. It also flags an overbroad unauthenticated authorization example that should be narrowed.
curl --proto '=https' --tlsv1.2 -sSfL https://run.linkerd.io/install | sh
The install template pipes a remote script from https://run.linkerd.io/install directly into sh. This executes network-delivered code without checksum or review, creating supply-chain and command-execution risk.
The ServerAuthorization template allows unauthenticated clients from 10.0.0.0/8. This can teach overly broad ingress trust if copied without narrowing networks or using meshTLS identities.
The unauthenticated setting and broad private CIDR are explicit in the policy example. The text frames it as ingress traffic, so legitimate use is possible but it needs stronger qualification.
Capability review items (1)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
curl --proto '=https' --tlsv1.2 -sSfL https://run.linkerd.io/install | sh
The URL retrieves the Linkerd installer during the installation workflow. It is a legitimate project URL, but it still introduces network-sourced code into the setup path.
Static analysis flagged shell, network, and weak-crypto patterns in SKILL.md. Review found these are documentation examples for Linkerd and Kubernetes, with one real operational risk from a curl-to-shell installer and multiple cluster-mutating commands.
1
Files scanned
310
Lines analyzed
5
Review items
1
False positives ignored
Capability review items (3)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
SKILL.md documents downloading the Linkerd installer from a URL and piping it directly to sh. This is a real supply-chain risk if copied without verification, although it appears in a legitimate installation guide and not in executable skill code.
The pipe-to-shell command is explicit on line 58. Context shows documentation rather than hidden execution, so the risk is operational guidance, not confirmed malware.
The skill includes Linkerd and kubectl commands that install CRDs, install the control plane, apply multi-cluster resources, and label services. These examples are expected for the skill purpose but can change live Kubernetes clusters if run without review.
The commands are clearly shown in bash blocks and perform Kubernetes changes. Their legitimate Linkerd context lowers concern about malicious intent.
The skill links to Linkerd documentation and an example API server address. These network references support user education and do not transmit local data.
The URLs are visible documentation or placeholder endpoint references. No code sends credentials, files, or environment data to those URLs.
Static false positives ignored (1)
These static matches were dismissed by semantic review or matched schema-only tokens, so they are shown for transparency but do not drive the quality score.
Several external command detections occur on Markdown code-fence boundaries or YAML examples, and the weak-crypto detections are words such as security-focused and identity. The hardcoded IP finding is a private CIDR in an authorization policy example.
Line review shows these signals are prose, diagrams, or Kubernetes YAML examples. They do not contain executable skill logic or evidence of secret exfiltration.
Static analysis flagged shell, network, and weak-crypto patterns in SKILL.md. Review found these are documentation examples for Linkerd and Kubernetes, with one real operational risk from a curl-to-shell installer and multiple cluster-mutating commands.
1
Files scanned
310
Lines analyzed
5
Review items
1
False positives ignored
Capability review items (3)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
SKILL.md documents downloading the Linkerd installer from a URL and piping it directly to sh. This is a real supply-chain risk if copied without verification, although it appears in a legitimate installation guide and not in executable skill code.
The pipe-to-shell command is explicit on line 58. Context shows documentation rather than hidden execution, so the risk is operational guidance, not confirmed malware.
The skill includes Linkerd and kubectl commands that install CRDs, install the control plane, apply multi-cluster resources, and label services. These examples are expected for the skill purpose but can change live Kubernetes clusters if run without review.
The commands are clearly shown in bash blocks and perform Kubernetes changes. Their legitimate Linkerd context lowers concern about malicious intent.
The skill links to Linkerd documentation and an example API server address. These network references support user education and do not transmit local data.
The URLs are visible documentation or placeholder endpoint references. No code sends credentials, files, or environment data to those URLs.
Static false positives ignored (1)
These static matches were dismissed by semantic review or matched schema-only tokens, so they are shown for transparency but do not drive the quality score.
Several external command detections occur on Markdown code-fence boundaries or YAML examples, and the weak-crypto detections are words such as security-focused and identity. The hardcoded IP finding is a private CIDR in an authorization policy example.
Line review shows these signals are prose, diagrams, or Kubernetes YAML examples. They do not contain executable skill logic or evidence of secret exfiltration.
All static findings evaluated as false positives. The skill contains only documentation and YAML templates for Linkerd service mesh. Patterns detected (URLs, shell pipes, network CIDRs) are legitimate infrastructure documentation. No actual code execution, cryptographic operations, or malicious patterns present.
This is a pure documentation skill with no executable code. It contains only markdown templates, YAML configurations, and bash command examples for educational purposes. No scripts, network calls, or file system access beyond its own directory.
4
Files scanned
549
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.
This is a pure documentation skill with no executable code. It contains only markdown templates, YAML configurations, and bash command examples for educational purposes. No scripts, network calls, or file system access beyond its own directory.
4
Files scanned
549
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.