Back to skill

Security audit

deep-coding-p

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed multi-agent coding workflow, but its local dashboard and broad agent permissions create review-worthy exposure of project files and logs.

Install only in a dedicated project workspace or container, keep the dashboard bound to localhost, do not keep secrets or credentials in that workspace, and stop preview/dashboard servers when done. Before wider use, fix the logs path traversal, restrict file-serving to an allowlist, sanitize rendered markdown, and pin or vendor frontend and npx dependencies.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Findings (3)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
assets/server.py:135
Finding

Directory Traversal in the Project Logs API

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
assets/server.py:229
Finding

Unauthenticated Static-File Access Exposes the Entire Workspace

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
assets/dashboard.html:7
Finding

Unpinned Third-Party JavaScript Executes in the Dashboard Origin

Content
View full analysis
``` ### Technical Analysis The dashboard downloads and executes JavaScript from three external CDN providers. The references use broad major-version ranges or no version at all: - Tailwind CSS is loaded from an unversioned CDN URL. - Vue is constrained only to major version 3. - Day.js is constrained only to major version 1. - markdown-it is constrained only to major version 13. None of the script tags includes a Subresource Integrity hash. The effective code executed by the dashboard can therefore change after the skill package has been reviewed. Remote scripts run with the privileges of the dashboard page. They can call same-origin endpoints, access loaded project state and logs, manipulate links, and make outbound network requests. A CDN compromise, package publishing compromise, malicious replacement, or incompatible upstream update could consequently affect every user who opens the dashboard. ### Attack Path 1. The user starts the local dashboard and opens it in a browser. 2. The page requests JavaScript from the configured CDN URLs. 3. A CDN response or mutable package release is compromised or unexpectedly changed. 4. The browser executes the changed script in the dashboard page. 5. The script accesses same-origin project APIs or workspace files exposed by the server. 6. The script can display falsified information or transmit accessible data to an external endpoint. ### Impact Assessment Successful supply-chain compromise would permit arbitra ...[truncated 687 chars]
Remediation
View remediation
``` 4. Generate and review a lockfile or dependency manifest for all vendored assets. 5. Add a restrictive Content Security Policy that permits scripts only from explicitly approved locations and limits outbound connections, for example: ```text default-src 'self'; script-src 'self'; connect-src 'self'; object-src 'none'; base-uri 'none'; frame-ancestors 'none' ``` 6. Remove runtime CDN compilation where possible, particularly the Tailwind CDN script, and use precompiled local CSS. 7. Establish a controlled dependency-update process with integrity verification and security review before changing bundled versions. ]]>
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (23)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description suggests a core capability centered on active multi-agent software engineering workflows (orchestration, code building, review, and iterative collaboration). The supplied code does not implement agent coordination, code generation, module decomposition, or review logic. Instead, it is a local dashboard/web server for viewing project states and logs and serving related files. This is a materially different primary purpose, so the description does not accurately represent the code chunk.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 253)May include surrounding context.

md
Dashboard is included in `assets/server.py` and `assets/dashboard.html`.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 256)May include surrounding context.

md
Dashboard is included in `assets/server.py` and `assets/dashboard.html`.

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 65)May include surrounding context.

html
</head>
<body class="bg-gray-50 min-h-screen">
    <div id="app" v-cloak>
        <!-- Header -->
        <header class="bg-white shadow-sm border-b">
            <div class="max-w-7xl mx-auto px-4 py-4">
                <div class="flex items-center justify-between">

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 169)May include surrounding context.

html
</div>
            </div>

            <!-- Pending Request Banner -->
            <div v-if="currentProject && currentProject.is_request" class="bg-gradient-to-r from-yellow-500 to-orange-500 rounded-lg shadow-lg p-6 mb-6 text-white">
                <div>
                    <h2 class="text-2xl font-bold">⏳ {{ currentProject.name.replace('[待启动] ', '') }}</h2>

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 283)May include surrounding context.

html
</div>
            </div>

            <!-- Task List -->
            <div v-if="currentProject && !currentProject.is_request" class="bg-white rounded-lg shadow">
                <div class="px-4 py-3 border-b border-gray-200">
                    <h2 class="text-lg font-bold text-gray-900">任务列表</h2>

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 320)May include surrounding context.

html
</div>
                                </div>
                                
                                <!-- Progress Info -->
                                <div v-if="module.progress" class="flex items-center gap-4 text-xs text-gray-500">
                                    <span>迭代:{{ module.progress.review_iterations || 0 }} 轮</span>
                                    <span v-if="module.progress.started_at">

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 320)May include surrounding context.

html
</div>
                                </div>
                                
                                <!-- Progress Info -->
                                <div v-if="module.progress" class="flex items-center gap-4 text-xs text-gray-500">
                                    <span>迭代:{{ module.progress.review_iterations || 0 }} 轮</span>
                                    <span v-if="module.progress.started_at">

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 382)May include surrounding context.

html
<div v-for="(it, index) in module.review_history" :key="index"
                                         class="rounded-lg overflow-hidden border"
                                         :class="getIterationBorderClass(it.status)">
                                        <!-- Header bar -->
                                        <div class="flex items-center justify-between px-3 py-2"
                                             :class="getIterationBgClass(it.status)">
                                            <div class="flex items-center gap-2">

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 403)May include surrounding context.

html
<!-- Detail panel -->
                                        <div v-if="isIterationExpanded(module.module_id, it.iteration)" class="px-3 py-3 bg-white space-y-3">
                                            
                                            <!-- Parsed checklist -->
                                            <div v-if="getParsedChecks(it.summary).length > 0">
                                                <p class="text-xs font-medium text-gray-500 mb-1.5">检查项:</p>
                                                <div class="space-y-1">

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/dashboard.html (reported line 421)May include surrounding context.

html
<div v-for="(line, li) in getNonCheckLines(it.summary)" :key="li">{{ line }}</div>
                                            </div>
                                            
                                            <!-- Feedback (only if different from summary) -->
                                            <div v-if="it.feedback && !isSameAsSummary(it.summary, it.feedback)" class="pt-3 border-t border-gray-100">
                                                <p class="text-xs font-medium text-gray-500 mb-1">反馈:</p>
                                                <div class="text-xs text-gray-600 space-y-1 leading-relaxed" v-html="renderMarkdown(it.feedback)"></div>

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill instructs users to run a local dashboard server, read and write project files, and use agent/session capabilities, yet it declares no explicit tool scope or permissions boundary. That mismatch can cause overbroad tool access at install/runtime, making it easier for the skill to read sensitive files or invoke unintended capabilities without clear user review.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

Using npx playwright without a pinned version allows whatever version is current at execution time to be fetched and run, which weakens supply-chain integrity and reproducibility. A compromised or breaking upstream release could introduce malicious code execution or unsafe behavior into the review workflow.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This HTML file declares lang="zh-CN", and the visible UI strings throughout the page are presented only in Chinese. Under the policy rule for natural-language violations, forcing a specific language/locale without user opt-in or justification is a reportable issue.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The dashboard renders agent-supplied markdown via markdown-it and injects the resulting HTML with Vue's v-html, but there is no sanitization step. Because review feedback and handoff summaries are attacker-controlled in this threat model, malicious HTML or script-capable payloads can execute in the operator's browser, turning a passive monitoring UI into an XSS sink.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The manifest describes a multi-agent development system for decomposing modules, implementing code, and reviewing quality, but this file adds a local HTTP API that enumerates and serves agent log contents from project directories. Hosting a dashboard/log-serving interface is not an obvious or declared requirement of the stated coding/orchestration purpose, especially because it exposes internal execution data over an HTTP surface.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The code serves files from the skill base directory, including arbitrary paths under /projects/, /archive/, and other static files under BASE_DIR, turning the skill into a local file-hosting web server. A deep-coding multi-agent skill may reasonably read project state internally, but publishing those files through HTTP is a separate capability not justified by the manifest's stated purpose.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The manifest describes a multi-agent development system for decomposing projects, implementing code, and reviewing quality. These instructions go beyond code review and orchestration by explicitly directing agents to start HTTP servers (python3 -m http.server or npx serve) and interact with running services, which is an operational capability not declared in the stated purpose.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The delivery workflow instructs the orchestrator to start and keep a local web server running, publish a localhost URL, and delay cleanup until user confirmation or timeout. That expands the skill from build/review orchestration into persistent service operation, increasing attack surface and creating opportunities for unintended data exposure, stale services, or misuse of local resources.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The delivery workflow defines accepted user responses using Chinese phrases alongside a couple of English terms, but provides no explicit language choice or opt-in. This can bias the skill toward a specific locale for completion handling and may exclude users communicating in other languages.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This code prepends the Chinese label "[待启动]" to request names for display, which imposes a specific language in the UI. The file does not indicate any user opt-in, locale selection, or justification for restricting output to Chinese.

Content

No source excerpt is available for this finding.