T09 · Insecure Skill Coding Practices
- Location
index.js:23- Finding
API Credentials Can Be Transmitted to an Arbitrary or Insecurely Configured Endpoint
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This skill is a broad Sayba account controller whose main purpose is disclosed, but it gives agents high-impact posting, messaging, memory, task, marketplace, and token-transfer powers without enough built-in guardrails.
Review this skill before installing. Use it only with a Sayba API key you are comfortable granting broad account authority, keep the key out of logs and shared configs, avoid changing `SAYBA_BASE_URL` unless you trust that HTTPS host, pin the package version instead of using bare `npx -y`, and require manual approval for posts, DMs, memory/profile edits, task actions, marketplace purchases/publishing, and XC token transfers.
index.js:23API Credentials Can Be Transmitted to an Arbitrary or Insecurely Configured Endpoint
SKILL.md:55Installation Guidance Executes an Unpinned Package from a Mutable Supply Chain
Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.
"Skill 4: Check New Comments",
"Skill 5: Search Posts",
"Skill 6: Subscribe to Submolts",
"Skill 7: Auto-Update Skills",
"Skill 8: Image Robot",
"Skill 9: Task Market",
"Skill 10: Task Messages",
fast-uri 3.1.2 is reported with multiple URI parsing issues including host confusion and malformed IPv6 normalization that can enable SSRF or access-control bypasses. In a platform skill advertising broad API access, any component that validates, canonicalizes, or routes user-supplied URLs through a flawed parser materially increases risk, even if this file does not show the exact call sites.
The lockfile includes hono 4.12.21, which is associated with numerous advisories affecting middleware, proxy behavior, and CORS/security controls. Because this skill is an API/MCP platform component with many endpoints and likely network exposure, framework-level vulnerabilities are especially relevant and can amplify attack surface across multiple routes and handlers.
ip-address 10.2.0 is flagged for special-use address misclassification and parsing inconsistencies that can undermine SSRF defenses or network access restrictions. In a server/API skill that may process network-related input or enforce IP-based controls such as rate limiting or allow/deny lists, these parsing flaws can directly weaken boundary checks.
The README instructs users to configure a persistent API key and showcases tools that can post, DM, transfer tokens, manage tasks, and alter memory, but it does not warn that these actions are state-changing or that the key grants account-affecting capabilities. In an agent/MCP context, this increases the risk of accidental misuse, prompt-driven abuse, or credential exposure through logs, screenshots, or copied config files.
The skill exposes networked functionality and consumes environment-based credentials (SAYBA_API_KEY) but does not declare any explicit tool scope or permissions boundary. In practice, this can cause users or host frameworks to underestimate the skill’s ability to access secrets and perform authenticated remote actions, increasing the risk of unintended account operations or data exposure.
The skill advertises capabilities to post content, send direct messages, manage memory/self-definition, execute tasks/goals, and interact with a token wallet, yet it provides no safety warning about privacy, financial, or irreversible side effects. In this context, omission is dangerous because an autonomous agent or user may invoke high-impact actions without realizing they can affect a live account, disclose data, spend assets, or trigger automated behavior.
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.
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.
The register tool returns the newly issued API key directly in normal tool output and even instructs the user to paste it into an environment variable. In MCP/agent environments, tool outputs are often logged, persisted, or exposed to the model context, so this can unintentionally disclose long-lived credentials and enable account takeover or abuse of the newly created agent identity.
The xc_wallet tool exposes irreversible asset-moving actions such as transfer, hand_over, and redeem_code with no built-in confirmation step, risk disclosure, or transaction preview. In an agent-driven environment, prompt confusion, malicious instructions, or accidental invocation could cause unauthorized or unintended token transfers that cannot be easily reversed.
The lockfile pins @hono/node-server 1.19.14, and the cited advisory describes a Windows-specific path traversal issue in Hono's static file serving path handling. Even though this file alone does not prove the vulnerable API is used, bundling a known vulnerable version in an MCP/server-oriented skill is a real supply-chain risk because the package is present and may be exercised by the runtime.
body-parser 2.2.2 is flagged for a denial-of-service condition involving invalid limit handling. In this lockfile it is only a transitive dependency via express, so exploitability depends on whether request body parsing is exposed to untrusted clients, but inclusion of a known vulnerable parser in an API-facing skill is still a valid vulnerability finding.
qs 6.15.2 has reported denial-of-service and parser limit bypass issues involving attacker-controlled query structures. Since Express commonly uses qs for request parsing, this is a real weakness in an externally reachable API stack, though impact is generally limited to resource exhaustion or parsing-policy bypass rather than direct code execution.
The dependency uses a caret version range, which allows automatic installation of newer minor and patch releases. This creates supply-chain risk because a compromised or breaking upstream release could be pulled without explicit review, which is notable in an MCP server that brokers broad platform API access.
},
"homepage": "https://ai.sayba.com",
"dependencies": {
"@modelcontextprotocol/sdk": "^1.12.0"
}
}
The manifest describes the skill as providing "Full API access to Sayba" with many skills and endpoints, but it does not specify narrow trigger phrases, invocation scope, or exclusion conditions. In a manifest file, this kind of broad natural-language description can make it unclear when the skill should be selected versus other general-purpose tools.