Back to skill

Security audit

TaskSquad.ai

Security checks for vulnerabilities and agentic risk

Overview

TaskSquad is a coherent remote-agent collaboration skill, but it asks users to install and run a daemon that can execute remotely assigned local tasks with under-scoped safety guidance.

Review before installing. Only use this with a TaskSquad account and agents you trust, avoid curl-to-bash installation unless you can verify the installer, treat tokens as secrets, run the daemon as a low-privilege user in a limited working directory, and assume task logs and live streams may reveal sensitive local agent output.

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)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:24
Finding
Unverified Remote Script Downloaded and Executed by Bash<![CDATA[ ## Vulnerability Details **File Location**: `SKILL.md`, lines 24-28 **Vulnerability Type**: Remote payload retrieval and immediate execution **Risk Level**: High ```markdown Using installation script (macOS/Linux/Windows): ``` curl -sSL install.tasksquad.ai | bash ``` ``` ### Technical Analysis The installation instructions pipe a remotely downloaded script directly into Bash. The payload is mutable and is neither pinned to a specific version nor verified with a cryptographic checksum or digital signature. The `-sSL` options also suppress normal curl output and follow redirects, allowing the final script to originate from a redirected location without being visibly reviewed. HTTPS protects the connection in transit but does not guarantee that the server, DNS configuration, deployment pipeline, domain account, or published script remains trustworthy. Because the downloaded installer is not included in the audited project, its behavior and effective permissions cannot be statically inspected. Installing a CLI may be necessary for the declared TaskSquad integration, but immediate execution of unverified remote content exceeds the minimum mechanism necessary to perform that installation. ### Attack Path 1. An attacker compromises `install.tasksquad.ai`, its DNS, hosting account, deployment pipeline, or a redirect destination. 2. The attacker replaces the legitimate installer with a malicious shell script. 3. A user or AI agent follows the documented installation command. 4. Curl downloads the attacker-controlled response and streams it directly to Bash. 5. Bash executes the payload immediately without an inspection or integrity-verification step. 6. The payload acts with the privileges of the invoking account and can download further components, access available credentials, alter files, or establish persistence. ### Impact Assessment Successful exploitation provides arbitrary command execution with the privileges of the user running the command. ...[truncated 599 chars]
Remediation
<![CDATA[ ## Remediation Suggestions 1. Remove the `curl | bash` installation method. 2. Publish immutable, versioned release artifacts through a documented official release channel. 3. Pin installation instructions to an exact release version rather than a mutable endpoint. 4. Publish SHA-256 or stronger checksums over a separately authenticated channel. 5. Sign releases and document signature verification using a clearly identified signing key. 6. Require users to download the artifact, verify its checksum and signature, inspect it where appropriate, and only then execute or install it. 7. Document the files, services, network connections, and permissions created by the installer. 8. Ensure installation does not require administrator privileges unless a specific component demonstrably needs them. 9. If a shell installer must remain available, host each version at an immutable URL and provide a safe workflow such as: ```bash curl --fail --proto '=https' --tlsv1.2 -o install-tsq.sh \ 'https://example.invalid/releases/vX.Y.Z/install-tsq.sh' echo '<EXPECTED_SHA256> install-tsq.sh' | sha256sum --check less install-tsq.sh bash install-tsq.sh ``` 10. Treat checksum placeholders as documentation examples only; publish a real version-specific digest before recommending the command. ]]>

T08 · Insecure Dependencies

Warning
Location
SKILL.md:18
Finding
Unpinned CLI Installation from a Third-Party Homebrew Tap<![CDATA[ ## Vulnerability Details **File Location**: `SKILL.md`, lines 18-22 **Vulnerability Type**: Unpinned third-party dependency and supply-chain exposure **Risk Level**: Medium ```markdown Using Homebrew (macOS/Linux): ``` brew tap xajik/tap && brew install tsq ``` ``` ### Technical Analysis The installation command adds the third-party Homebrew tap `xajik/tap` and installs `tsq` without pinning a formula revision or release version. The command therefore trusts the current and future state of an external repository that is not included in the audited project. A compromised repository owner account, malicious maintainer update, altered release asset, or compromised upstream build process could cause Homebrew to retrieve and execute attacker-controlled installation logic. The Skill provides no checksum, signature, commit pin, provenance statement, or reproducible-build verification. Using a package manager is a reasonable way to install the declared CLI, but relying on an unpinned third-party source without documented integrity controls creates avoidable supply-chain risk. ### Attack Path 1. An attacker compromises the third-party tap repository, a maintainer account, an upstream release location, or its build and publication pipeline. 2. The attacker modifies the `tsq` formula or referenced release artifact. 3. A user follows the documented `brew tap` and `brew install` command, or later upgrades the package. 4. Homebrew retrieves the modified formula or artifact. 5. Malicious installation logic or a malicious `tsq` binary executes under the invoking user's account. 6. The attacker can access user-readable data, modify developer tooling or Homebrew-managed content, and potentially retain execution through the installed binary or related configuration. ### Impact Assessment Exploitation could provide arbitrary code execution with the privileges of the account running Homebrew. The affected scope may include source repositories, environment variabl ...[truncated 455 chars]
Remediation
<![CDATA[ ## Remediation Suggestions 1. Prefer a verified first-party package repository or an established distribution channel with documented ownership. 2. Pin the CLI to an exact reviewed release version and, where practical, pin the tap to a reviewed commit. 3. Publish and verify cryptographic checksums and release signatures for the formula and referenced artifacts. 4. Enable release provenance controls, protected branches, mandatory review, signed commits or tags, and strong multifactor authentication for maintainers. 5. Document the tap owner, source repository, release process, and expected artifact hashes so users can validate provenance. 6. Avoid unattended upgrades for this daemon until new versions have been reviewed and verified. 7. Publish a software bill of materials and reproducible-build instructions where feasible. 8. Run the daemon as a dedicated, unprivileged account with access limited to explicitly approved working directories and credentials. 9. Require authorization and, for sensitive operations, explicit approval before remotely assigned tasks can launch local commands. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (19)

Missing User Warnings

High
Confidence
99% confidence
Finding
The installation instructions recommend piping a remotely fetched script directly into bash without any integrity verification, review step, or safety warning. This is dangerous because a compromised install host, DNS/TLS issue, or malicious script update would immediately execute arbitrary code on the user's system.

External Script Fetching

High
Category
Supply Chain
Content
Using installation script (macOS/Linux/Windows):

```
curl -sSL install.tasksquad.ai | bash
```

### Prerequisite: tmux
Confidence
99% confidence
Finding
Fetching an install script from a remote host and executing it immediately is a classic unsafe pattern because it grants arbitrary code execution to the remote source at install time. Users are given no opportunity to verify script contents, origin integrity, or version, making supply-chain compromise especially impactful.

Chaining Abuse

High
Category
Tool Misuse
Content
Using installation script (macOS/Linux/Windows):

```
curl -sSL install.tasksquad.ai | bash
```

### Prerequisite: tmux
Confidence
98% confidence
Finding
The explicit shell chaining operator into bash converts remote content retrieval directly into command execution, removing any review boundary between download and run. In a skill that already enables remote agent orchestration, normalizing this pattern increases the chance of users accepting unsafe execution flows and suffering full host compromise if the upstream source is malicious.

Missing User Warnings

High
Confidence
98% confidence
Finding
The skill states that the daemon pulls tasks from a central portal and that the agent executes them locally, but it provides no explicit warning that remote task content can cause code or command execution on the user's machine. In this context, omitting that warning is dangerous because the core feature is remote-to-local task execution, which materially increases risk of arbitrary action, data access, or system compromise if tasks are malicious or misrouted.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill tells users to obtain a token from the browser but gives no warning that the token is a sensitive credential or guidance on secure handling. This is dangerous because users may paste the token into logs, shell history, shared terminals, or unsecured files, enabling account or agent takeover.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
60% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Create a new team in your account.

```bash
curl -X POST "https://api.tasksquad.ai/teams" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Team"}'
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The live-streaming and task-log features are presented without warning that they may expose agent output, prompts, filesystem paths, secrets, or other sensitive operational data to remote viewers. In an agent-execution skill, such telemetry can leak confidential information well beyond what users expect.

Static analysis

No suspicious patterns detected.