Audit History

assets/github_actions_template.yml:24-27 assets/github_actions_template.yml:35-37

Pipe to shell pattern

curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b /usr

The CI template downloads an installer from the mutable master branch and immediately executes it with sh. A compromised upstream response would gain code execution in the runner.

High

Mutable CI Dependencies Execute in a Privileged Workflow

The GitHub workflow uses a mutable major action tag, requests the latest reviewdog release, and installs unpinned Python packages. Upstream changes can execute before later steps receive a write-capable GitHub token.

The mutable action tag, latest version selector, and unpinned package installation are explicit. The workflow also grants pull-request and checks write permissions.

High

assets/.reviewdog.yml:38-46 assets/github_actions_template.yml:79-99 assets/gitlab_ci_template.yml:67-86 assets/pre_commit_config.yaml:37-45

Critical Secret Scans Fail Open

Gitleaks failures are suppressed with true, and missing reports can produce empty findings or skip reviewdog. Scanner crashes can therefore pass the critical secret gate.

Each configuration explicitly suppresses the Gitleaks exit status. At least one path substitutes empty findings or skips review processing when output is absent.

Capability review items (4)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

Hardcoded URL

curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b /usr

This URL targets a mutable master-branch installer that is fetched during CI and passed directly to sh. The network dependency is not pinned or integrity-verified.

Medium

assets/.reviewdog.yml:41

Temp directory access

gitleaks detect --report-format json --report-path /tmp/gitleaks.json --no-git 2>/dev/null || true

Gitleaks writes sensitive scan output to the predictable shared path /tmp/gitleaks.json. On a shared host, collisions or symlink attacks could overwrite another file or expose findings.

Medium

assets/.reviewdog.yml:42

Temp directory access

cat /tmp/gitleaks.json 2>/dev/null || echo '{"findings":[]}'

The runner reads scan results from the predictable shared path /tmp/gitleaks.json. A colliding or attacker-controlled file could be accepted as trusted scanner output.

Medium

assets/github_actions_template.yml:43 assets/github_actions_template.yml:56 assets/github_actions_template.yml:70 assets/github_actions_template.yml:82 assets/github_actions_template.yml:105 assets/github_actions_template.yml:121 assets/github_actions_template.yml:139 references/cwe_mapping.md:304 references/cwe_mapping.md:331 references/reporter_formats.md:46 references/reporter_formats.md:62 references/reporter_formats.md:159 references/reporter_formats.md:403 references/reporter_formats.md:417 SKILL.md:63 SKILL.md:116 SKILL.md:313

Temp directory access

entry: bash -c 'gitleaks detect --report-format json --report-path /tmp/gitleaks.json --no-git 2>/de

The pre-commit hook writes and reads Gitleaks results at the fixed shared path /tmp/gitleaks.json. This permits collisions and symlink manipulation on multi-user systems.

Risk Factors

📁 Filesystem access (34)

🔑 Env variables (17)

🌐 Network access (19)

assets/gitlab_ci_template.yml:14 assets/pre_commit_config.yaml:69 assets/pre_commit_config.yaml:81 assets/pre_commit_config.yaml:88 references/reporter_formats.md:71 references/reporter_formats.md:139 references/reporter_formats.md:245 references/reporter_formats.md:279 references/reporter_formats.md:295 references/supported_tools.md:421 references/supported_tools.md:435 SKILL.md:6 SKILL.md:20 SKILL.md:21 SKILL.md:381 SKILL.md:382 SKILL.md:383 SKILL.md:384 SKILL.md:385

⚡ Contains scripts (2)

references/reporter_formats.md:331 SKILL.md:45-54 SKILL.md:54-58 SKILL.md:58-66 SKILL.md:66-74 SKILL.md:74-83 SKILL.md:83-101 SKILL.md:101-125 SKILL.md:125-128 SKILL.md:128-139 SKILL.md:139-145 SKILL.md:145-157 SKILL.md:157-178 SKILL.md:178-191 SKILL.md:191-192 SKILL.md:192-197 SKILL.md:197-199 SKILL.md:199-200 SKILL.md:200-202 SKILL.md:202-204 SKILL.md:204-205 SKILL.md:205-206 SKILL.md:206-208 SKILL.md:208-210 SKILL.md:210-211 SKILL.md:211-212 SKILL.md:212-213 SKILL.md:213-221 SKILL.md:221-235 SKILL.md:235-241 SKILL.md:241-251 SKILL.md:251-257 SKILL.md:257-263 SKILL.md:263-269 SKILL.md:269-277 SKILL.md:277-312 SKILL.md:312 SKILL.md:313 SKILL.md:320-321 SKILL.md:321-322 SKILL.md:322-328 SKILL.md:328-337 SKILL.md:337-339 SKILL.md:339-343 SKILL.md:343-345 SKILL.md:345-364 SKILL.md:364-370 SKILL.md:370-377 SKILL.md:222

⚙️ External commands (49)

Detected Patterns

Pipe to shell pattern

Audited by: codex

Audit version 8

Jul 7, 2026, 09:33 PM

The skill is primarily documentation and CI templates for reviewdog-based security scanning. Most static findings are expected scanner commands, CI token references, or security reference examples. The GitLab CI template contains a confirmed pipe-to-shell remote installer, and fixed /tmp Gitleaks paths should be replaced.

Files scanned

2,206

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (1)

Critical

Pipe to shell pattern

curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b /usr

The GitLab template downloads install.sh from the reviewdog main branch and pipes it directly to sh. That executes unauthenticated network content in CI and is a real supply chain risk.

Capability review items (4)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

Hardcoded URL

curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b /usr

The hardcoded raw.githubusercontent.com URL is part of a remote installer that is executed by the shell. The endpoint is legitimate, but the unpinned network fetch materially increases CI supply chain risk.

Medium

assets/.reviewdog.yml:41

Temp directory access

gitleaks detect --report-format json --report-path /tmp/gitleaks.json --no-git 2>/dev/null || true

The sample writes Gitleaks output to a predictable /tmp/gitleaks.json path. In shared or reused environments this can collide with other processes or expose sensitive finding metadata.

Medium

assets/.reviewdog.yml:42

Temp directory access

cat /tmp/gitleaks.json 2>/dev/null || echo '{"findings":[]}'

The sample reads a predictable /tmp/gitleaks.json path after scanner execution. That fixed temp path can consume stale or attacker-created content in shared environments.

Medium

Temp directory access

entry: bash -c 'gitleaks detect --report-format json --report-path /tmp/gitleaks.json --no-git 2>/de

The pre-commit hook writes and reads /tmp/gitleaks.json using a fixed name. A per-run temporary file would avoid collisions and stale secret-scan output.

Risk Factors

📁 Filesystem access (34)

🔑 Env variables (17)

🌐 Network access (19)

⚡ Contains scripts (2)

⚙️ External commands (49)

Detected Patterns

Pipe to shell pattern

Audited by: codex

Audit version 7

Jul 6, 2026, 01:08 AM

I found no evidence of prompt injection or malicious data exfiltration in the skill text. Most static alerts are safe documentation examples or expected reviewdog CI configuration. Confirmed risks are the GitLab CI remote installer pipe-to-shell pattern and a predictable temporary Gitleaks report path.

Files scanned

2,087

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (1)

Critical

Pipe to shell pattern

curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b /usr

The GitLab CI template downloads an installer from a mutable raw GitHub branch and executes it with sh. This creates a real supply-chain execution risk without pinning or checksum verification.

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.

Medium

Standard device file access

entry: bash -c 'gitleaks detect --report-format json --report-path /tmp/gitleaks.json --no-git 2>/de

The hook writes Gitleaks output to a predictable /tmp/gitleaks.json path. Secret scan reports can contain sensitive values, so this can leak data or collide with another local process.

Medium

Temp directory access

entry: bash -c 'gitleaks detect --report-format json --report-path /tmp/gitleaks.json --no-git 2>/de

The hook stores a Gitleaks report in a fixed temporary path under /tmp. Because these reports may include secrets, the file should use a private temporary file and be deleted after use.

Low

Hardcoded URL

curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b /usr

The hardcoded URL is used to fetch a remote installer during CI. The risk is supply-chain exposure from an external mutable dependency, especially because the script is executed immediately.

Risk Factors

📁 Filesystem access (21)

🔑 Env variables (17)

🌐 Network access (19)

⚡ Contains scripts (2)

⚙️ External commands (49)

Detected Patterns

Pipe to shell pattern

Audited by: codex

Audit version 6

Jun 28, 2026, 06:01 AM

Static analysis found many command, network, filesystem, and token patterns. Most are expected for a reviewdog CI integration skill, but the GitLab template includes a confirmed curl-to-shell installer pattern that should be remediated before publication. No prompt injection or confirmed malicious exfiltration intent was found.

Files scanned

2,087

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (1)

High

SKILL.md:45 SKILL.md:54 assets/pre_commit_config.yaml:12 assets/pre_commit_config.yaml:23 assets/pre_commit_config.yaml:42

Remote Installer Piped Directly To Shell

The GitLab CI template downloads the reviewdog install script over the network and executes it directly with sh. This is a real supply chain risk because CI would execute remote script contents without pinning, checksum verification, or review.

The static pattern is confirmed in a CI template and directly combines network download with shell execution. The use is probably intended for installation, but the execution pattern is unsafe for reusable marketplace content.

Needs review findings (2)

These findings came from uncertain legacy audit verdicts, so they require review but are not counted as confirmed security issues.

Medium

Repository Token Use Requires Least Privilege Controls

The GitHub Actions examples pass repository tokens to reviewdog so it can post pull request comments and checks. This is legitimate for reviewdog, but users must keep permissions minimal and avoid exposing tokens to untrusted pull request code.

The token references are clearly present and required for reviewdog reporters. Risk depends on repository settings, pull request trust boundaries, and scanner behavior rather than malicious code in the skill.

Medium

Shell-Based Scanner Pipelines Run In User Repositories

The skill provides shell pipelines that run scanners and pass results to reviewdog. This matches the skill purpose, but users should review scanner arguments, pinned tool versions, and file write locations before adopting the templates.

The command execution patterns are real but primarily demonstrate normal reviewdog usage. They become risky if copied into CI without pinning tools or reviewing behavior on untrusted repositories.

Static false positives ignored (3)

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

references/cwe_mapping.md:21 references/cwe_mapping.md:83 references/cwe_mapping.md:112 references/supported_tools.md:17

Security Keyword Findings Are Documentation Examples

Static analysis flagged weak cryptography, eval, private key, and command injection terms in reference material. These references describe vulnerability classes and scanner coverage, not instructions to perform unsafe actions.

The surrounding context is CWE and supported-tool documentation. The flagged terms are examples of issues that scanners detect.

Low

SKILL.md:20-21 assets/pre_commit_config.yaml:69 assets/pre_commit_config.yaml:81 assets/pre_commit_config.yaml:88

Hardcoded URLs Point To Public Documentation And Tool Sources

The hardcoded URL findings are mainly public project, documentation, and pre-commit repository URLs. They do not show data exfiltration endpoints or suspicious callback behavior.

The URLs are visible configuration and documentation references. No evidence found that they receive secrets or repository data.

Low

assets/github_actions_template.yml:72 assets/github_actions_template.yml:128 assets/pre_commit_config.yaml:42 references/supported_tools.md:64

Filesystem Access Is Normal Scanner Input And Artifact Handling

Filesystem detections mostly reflect scanner access to repository files, Docker volume mounts, and temporary report files. This behavior is expected for SAST, secret scanning, and linter integrations.

The file paths are used to scan project contents or store local scan reports. No evidence found of unauthorized file collection or exfiltration.

Risk Factors

📁 Filesystem access (23)

🔑 Env variables (17)

🌐 Network access (29)

⚡ Contains scripts (2)