Back to skill

Security audit

Dev Productivity Bundle

Security checks for vulnerabilities and agentic risk

Overview

This developer-workflow skill is mostly coherent, but its setup asks users to run an unreviewed remote installer and install mutable add-on skills, so it should be reviewed before installation.

Review the remote repository and install.ps1 before running it, prefer a pinned signed release or checksum, and install only the dependency skills you trust at reviewed versions. Use narrowly scoped GitHub tokens, keep webhook secrets out of source control, and configure monitoring only for services you own or are authorized to check.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T08 · Insecure Dependencies

Error
Location
SKILL.md:59
Finding

Unpinned Third-Party Skills Create a Supply-Chain Risk

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 59-72
Vulnerability Type: Unpinned and unaudited third-party dependencies
Risk Level: High

powershell
py -m clawhub install ai-refactoring-assistant
py -m clawhub install coding-agent

py -m clawhub install gh-issues
py -m clawhub install api-dev

py -m clawhub install academic-writing-refiner
py -m clawhub install feishu-doc

py -m clawhub install healthcheck
py -m clawhub install cron

Technical Analysis

The installation instructions retrieve eight third-party Skills by package name without specifying exact versions, immutable content digests, trusted publishers, or integrity checks. Consequently, the installed content depends on the mutable state of the package registry at installation time.

None of these dependencies are included in the audited artifact, so their code, permissions, transitive dependencies, and runtime behavior cannot be verified through this project. A registry compromise, publisher-account compromise, malicious package update, package-name reassignment, or dependency-confusion event could cause different and potentially malicious content to be installed under a documented name.

This is particularly sensitive because the advertised Skills interact with source code, GitHub issues, documentation platforms, health checks, and scheduled operations. Such integrations may receive repository credentials, API tokens, source code, or access to deployment infrastructure.

Attack Path

  1. An attacker compromises the publisher account or registry entry for one of the named Skills, or publishes a malicious replacement under a confusing or reassigned name.
  2. The attacker releases a malicious package version containing credential theft, unauthorized tool calls, or destructive behavior.
  3. A user follows the documented installation command without specifying a trusted version or digest.
  4. The package manager resolves and ins ...[truncated 977 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin every Skill to an exact, reviewed version or, preferably, an immutable content digest.
  2. Document the expected registry origin and verified publisher identity for each dependency.
  3. Require cryptographic signature and checksum verification before installation.
  4. Maintain a lock file recording resolved versions, digests, and transitive dependencies.
  5. Review third-party Skill contents and permission requirements before approving updates.
  6. Run installed Skills in a sandbox with least-privilege filesystem, network, tool, and credential access.
  7. Expose repository and service credentials only when needed, using narrowly scoped and short-lived tokens.
  8. Add automated dependency monitoring and require security review for lock-file changes.

T03 · Remote Payload Retrieval and Execution

Error
Location
README.md:7
Finding

Mutable Remote Repository Is Cloned and Its Unreviewed Installer Is Executed

Content
View full analysis

Vulnerability Details

File Location: README.md, lines 7-9
Vulnerability Type: Remote payload retrieval and execution
Risk Level: High

powershell
git clone https://github.com/clawhub/dev-productivity-bundle.git
cd dev-productivity-bundle
./install.ps1

Technical Analysis

The setup procedure clones the current default branch of a remote repository and immediately runs install.ps1. No immutable commit, signed release, version tag, checksum, or signature is specified. Therefore, the effective installer can change after this artifact has been reviewed.

The audited artifact does not contain install.ps1, meaning the installer referenced by the documentation was not available for inspection. The artifact also omits the advertised config/ directory and several other documented files. A user following the instructions would therefore execute code obtained from outside the audited package.

This creates a remote-payload execution channel: control of the referenced repository, its default branch, or the user's connection to an insufficiently verified source could alter the PowerShell script that users execute.

Attack Path

  1. An attacker compromises the referenced repository, a maintainer account, or the default branch's update process.
  2. The attacker modifies or introduces install.ps1 with malicious PowerShell commands.
  3. A user runs the documented git clone command, retrieving the current attacker-controlled repository state.
  4. The user changes into the cloned directory and executes ./install.ps1 without verifying its commit, signature, or content.
  5. The script executes with the privileges of the invoking user and can perform any operation allowed to that account.

Impact Assessment

A malicious installer could read or modify files accessible to the invoking user, steal locally available credentials, install additional software, alter development configuration, or make arbitrar ...[truncated 447 chars]

Remediation
View remediation

Remediation Suggestions

  1. Include install.ps1 in the distributed artifact so it can be reviewed together with the documentation.
  2. Reference a signed release or immutable commit hash rather than the repository's mutable default branch.
  3. Publish a cryptographic checksum for the installer and require verification before execution.
  4. Sign the PowerShell script with a trusted code-signing certificate and document signature validation.
  5. Avoid a clone-and-execute workflow; distribute a versioned package through a trusted registry with integrity metadata.
  6. Display and review the installer before execution, and run it without administrative privileges unless elevation is strictly necessary.
  7. Add strict error handling and document every filesystem, network, credential, and configuration change performed by the installer.
  8. Include all referenced configuration templates and example files in the audited package to make installation reproducible.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (7)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The documented trigger phrases are very broad, natural-language commands like '查看Bug情况' and '服务状态' that could plausibly appear in ordinary conversation and unintentionally invoke the skill. Because this bundle exposes code review, bug tracking, documentation, and deployment-monitoring capabilities, accidental activation could disclose project information or cause unintended actions in a developer workflow.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This markdown file includes explicit setup instructions for GITHUB_TOKEN and GITHUB_WEBHOOK_SECRET, which are sensitive credentials. Under the missing-warning rule for markdown files, the description should warn users about protecting secrets, avoiding committing them to source control, and using secure secret storage.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 451)May include surrounding context.

md
services:
  - name: "user-service"
    url: "https://api.example.com/health"
    check_type: "http"

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file presents headings, instructions, user commands, and responses only in Chinese, which effectively forces a specific language for interaction. The policy allows fixed-language behavior only when the skill offers user opt-in or clearly documents a justified locale constraint, neither of which appears here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
74% confidence
Finding

All user-facing instructions and example commands in the file are presented only in Chinese, with no indication that other languages are supported or that Chinese is an optional locale. This may violate language-choice policy if the skill is intended for a broader audience and does not document a justified region-specific constraint.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The natural-language description and instructions are primarily in Chinese, which imposes a language choice on users without opt-in. The policy requires flagging language or locale constraints unless the skill offers a choice or clearly documents a justified regional limitation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The bundle advertises deployment monitoring, health checks, and CI/CD integration, all of which can affect privacy or system integrity by interacting with production infrastructure. The markdown presents these capabilities as benign features without any user-facing disclosure about production access, monitoring scope, or operational caution.

Content

No source excerpt is available for this finding.