📦

Audit History

distributed-tracing - 8 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v8 LatestJul 7, 2026, 07:35 AM 2 confirmed0No capability change
v7 Jul 7, 2026, 07:35 AM 2 confirmed0Contains scripts
v6 Jul 1, 2026, 12:15 AM 2 confirmed1 Contains scripts
v5 Jan 21, 2026, 07:17 PM No confirmed findings0No capability change
v4 Jan 17, 2026, 08:29 AM No confirmed findings0No capability change
v3 Jan 17, 2026, 08:29 AM No confirmed findings0Network accessContains scriptsExternal commands
v2 Jan 4, 2026, 05:08 PM No confirmed findings0No capability change
v1 Jan 4, 2026, 05:08 PM No confirmed findings0Baseline

Jul 7, 2026, 07:35 AM

The static findings are mostly false positives caused by Markdown code fences, language examples, and visible example URLs in SKILL.md. I found no prompt injection, covert command execution, credential collection, or data exfiltration intent. Two production safety issues remain around sensitive trace metadata and unpinned container images.

1
Files scanned
439
Lines analyzed
5
Review items
0
False positives ignored

Confirmed security concerns (2)

Medium
Sensitive Data in Trace Attributes
The skill shows recording database statements and recommends user_id tags in spans. Production trace stores can expose personal data or sensitive query details if copied without redaction.
The cited lines explicitly include db.statement and user_id as trace metadata. These fields are common sources of sensitive data leakage in observability systems.
Low
Unpinned Observability Container Images
The Docker Compose and Tempo examples use latest image tags. This creates nondeterministic deployments and can introduce unreviewed component changes.
Both cited lines use latest tags for infrastructure images. This is a clear supply-chain hardening issue, though not evidence of malicious intent.
Audited by: codex

Jul 7, 2026, 07:35 AM

The static findings are mostly false positives caused by Markdown code fences, language examples, and visible example URLs in SKILL.md. I found no prompt injection, covert command execution, credential collection, or data exfiltration intent. Two production safety issues remain around sensitive trace metadata and unpinned container images.

1
Files scanned
439
Lines analyzed
5
Review items
0
False positives ignored

Confirmed security concerns (2)

Medium
Sensitive Data in Trace Attributes
The skill shows recording database statements and recommends user_id tags in spans. Production trace stores can expose personal data or sensitive query details if copied without redaction.
The cited lines explicitly include db.statement and user_id as trace metadata. These fields are common sources of sensitive data leakage in observability systems.
Low
Unpinned Observability Container Images
The Docker Compose and Tempo examples use latest image tags. This creates nondeterministic deployments and can introduce unreviewed component changes.
Both cited lines use latest tags for infrastructure images. This is a clear supply-chain hardening issue, though not evidence of malicious intent.
Audited by: codex

Jul 1, 2026, 12:15 AM

Static analysis reported 53 issues, but the weak-cryptography, dynamic-import, and shell-backtick alerts are false positives from Markdown front matter, Go import syntax, code fences, and inline references. Confirmed risk is medium because users may copy Kubernetes deployment commands and tracing examples that contact remote or internal endpoints and may store sensitive trace metadata. No evidence of malware, prompt injection, credential exfiltration, or hidden executable code was found.

1
Files scanned
439
Lines analyzed
5
Review items
2
False positives ignored

Confirmed security concerns (2)

Medium
Trace Metadata May Expose Sensitive Data
The examples show recording database statements and recommend meaningful tags such as user_id. Trace backends can retain and expose these values, so users should redact sensitive attributes before production use.
The skill explicitly demonstrates db.statement and user_id-style trace attributes. This is common observability practice, but it creates a real privacy risk if values are not sanitized.
Low
Backtick Execution Alerts Are Markdown Fences
The Ruby or shell backtick findings are caused by Markdown code fences and inline reference paths. No evidence found of shell backtick execution in runnable skill code.
The flagged syntax appears in documentation formatting and code examples. The skill contains only SKILL.md, so these snippets are not automatically executed.
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.

Medium
Copyable Kubernetes Deployment Commands
The shell example creates a namespace, fetches a Jaeger operator manifest from GitHub, and applies a Jaeger custom resource. This is legitimate observability guidance, but copying it into a production cluster can modify infrastructure and trust a remote manifest without review.
The deployment commands are directly present in a Markdown bash block. They appear legitimate for the skill purpose, so the risk is operational rather than malicious.
Static false positives ignored (2)

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.

Low
Dynamic Import Alert Is a False Positive
The dynamic import finding points to a Go import list inside a Markdown code example. It is not JavaScript dynamic import syntax and is not executed by the skill.
The referenced lines are clearly a Go import block within fenced Markdown. There is no runtime loader or user-controlled import behavior.
Low
Weak Cryptography Alert Is a False Positive
The weak cryptographic algorithm finding points to the YAML description line. No cryptographic algorithm or hashing implementation appears at that location.
Line 3 is plain descriptive metadata about distributed tracing. The scanner likely matched text accidentally, and there is no crypto usage to evaluate there.

Detected Patterns

Remote Manifest Applied to KubernetesHTTP Collector and Service Endpoints
Audited by: codex

Jan 4, 2026, 05:08 PM

Documentation-only skill with no executable code. Contains only markdown guidance, code examples, and configuration templates for users to copy into their own applications. Zero attack surface - no data access, no network calls, no file operations, no command execution.

4
Files scanned
657
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.
Audited by: claude

Jan 4, 2026, 05:08 PM

Documentation-only skill with no executable code. Contains only markdown guidance, code examples, and configuration templates for users to copy into their own applications. Zero attack surface - no data access, no network calls, no file operations, no command execution.

4
Files scanned
657
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.
Audited by: claude