Back to skill

Security audit

taskfuel

Security checks for vulnerabilities and agentic risk

Overview

The skill is rated suspicious because its setup runs a remote installer directly, and scanners also report undisclosed endpoint ratings and feedback plus paid calls that can proceed without user approval.

Consider the verdict suspicious before installing. The main concerns are the unverified remote installer, extra endpoint ratings and feedback, and paid operations that may not require your approval. If you proceed, review the install and spending steps carefully and approve installations and paid calls explicitly. The scans covered the reported Markdown component, but this verdict does not establish that the skill is risk-free.

Vulnerability Patterns
  • 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
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Findings (8)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The commands rate API endpoints rather than search for services or execute any of the declared capabilities. Endpoint voting is not disclosed in the description, and these standalone rating actions are not shown as supporting a declared task. The endpoint names do not establish that searching or generation actually occurs.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The supplied chunk's sole action is reporting a discovery failure to Taskfuel. It does not search for weather services or retrieve weather data. Feedback submission is not declared, and the chunk shows no surrounding task execution that would establish it as a supporting implementation detail. Its unconditional submission also conflicts with the description's requirement for user approval before sending a message.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
99% confidence
Finding

The command downloads mutable remote content and executes it directly, without version pinning or independent integrity verification. HTTPS protects transport but does not prevent a compromised publisher or distribution system from supplying code that runs with the agent's local privileges.

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

0. Ensure the CLI is installed

sh
command -v taskfuel || curl -fsSL https://taskfuel.ai/install.sh | sh

If ~/.local/bin is not on PATH, the installer says so — follow its hint.

Chaining Abuse

High
Category
Tool Misuse
Confidence
98% confidence
Finding

The pipe connects network-controlled bytes directly to a shell interpreter, creating a concrete remote-code-execution trust boundary. This confirms the unsafe execution mechanism, not malicious intent, and duplicates the installer issue identified by SQP-2 and SC2 rather than representing a separate exploit.

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

0. Ensure the CLI is installed

sh
command -v taskfuel || curl -fsSL https://taskfuel.ai/install.sh | sh

If ~/.local/bin is not on PATH, the installer says so — follow its hint.

The CLI setup automatically executes a downloaded installer without warning the user about changes to their system.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

If the CLI is missing, the setup command immediately executes a remotely downloaded installer without a separate approval or review step. In an agent shell, this can modify the environment and access files or credentials with the agent's permissions; exploitation would require malicious installer content or compromise of its delivery path.

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 198)May include surrounding context.

After calling an endpoint, rate it with the same method and URL you called:

sh
taskfuel rate https://api.example.com/search --vote up
taskfuel rate https://api.example.com/generate --method POST --vote down

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 199)May include surrounding context.

sh
taskfuel rate https://api.example.com/search --vote up
taskfuel rate https://api.example.com/generate --method POST --vote down
  • Vote up if you got the expected/useful result; vote down if the

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The skill permits paid calls below its own thresholds without establishing that the user authorized spending, and it explicitly allows choosing TaskFuel when the user never requested it. Quotes, caps, and balance checks limit financial exposure but do not supply consent; the separate approval requirement for communications reduces, but does not eliminate, the risk.

Content

Scanner excerpt · SKILL.md (reported line 262)May include surrounding context.

md
A `spec_unavailable` refusal is transient: retry the quote shortly.
- **Ask before big spends, and ask once**: a single call over $1, or a task
  whose paid calls will add up to more than $5, needs the user's explicit OK
  first. Below those lines, call without asking: the quote and the gateway's
  hard cap already bound what a call can cost, and asking for every $0.10
  call is what users complain about. An approval sticks for the session:
  once the user has OK'd an endpoint at a quoted price, do not ask again for