T09 · Insecure Skill Coding Practices
- Location
SKILL.md:210- Finding
Untrusted External Activity URLs Are Rendered Without Host Allowlisting
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly a read-only Gate activity helper, but it uses Gate credentials and has under-scoped account-data and link-handling behavior that should be reviewed before installation.
Install only if you are comfortable giving this skill read-only Gate activity access through your local MCP session. Review whether broad activity requests should invoke it, and prefer a version that validates activity links against approved Gate HTTPS hosts and only calls 'My Activities' when you explicitly ask for account-specific information.
SKILL.md:210Untrusted External Activity URLs Are Rendered Without Host Allowlisting
references/mcp.md:21Authenticated User-Specific Endpoint Is Probed for Unrelated Public Queries
Referenced artifact was not completely inspected
- `SKILL.md` keeps routing and recommendation logic.
Line L21 states that non-English keywords must be translated to English before the API call. This is a natural-language policy concern because it imposes an English-language handling requirement without indicating user choice, opt-in, or a clearly documented locale-specific justification in this file.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
### Key Rules
1. **Fixed page_size=3**: All activity list queries use `page_size=3`, no pagination
2. **No Type Clarification**: For hot recommendation, call API directly without asking user
3. **Type Matching**: Case-insensitive matching for activity types
## Source
Skill grants unrestricted tool access without appropriate constraints. An agent with unfettered tool access can perform arbitrary actions including file modification, network requests, and code execution.
## General Rules
⚠️ STOP — You MUST read and strictly follow the shared runtime rules before proceeding.
Do NOT select or call any tool until all rules are read. These rules have the highest priority.
→ Read `./references/gate-runtime-rules.md`
- **Only call MCP tools explicitly listed in this skill.** Tools not documented here must NOT be called, even if they
exist in the MCP server.
The trigger phrases are broad and include open-ended wording like "etc.", which can cause the skill to activate for ambiguous user requests unrelated to the intended activity-center workflow. In an agent setting, unclear invocation boundaries increase the chance of inappropriate tool use and unintended access to account-linked activity data.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- Credentials Source: Local Gate MCP deployment (`GATE_API_KEY`, `GATE_API_SECRET`)
- API Key Required: Yes
- Permissions: Activity:Read
- Never ask the user to paste secrets into chat; rely on the configured MCP session only.
- API Key Provisioning Reference: https://www.gate.com/myaccount/profile/api-key/manage (create or rotate keys outside the chat when the local MCP setup requires them).
### Installation Check
Forcing translation of non-English input to English before search changes user-provided query semantics without consent and can produce incorrect matches or unintended searches. In a security-sensitive assistant, silent transformation of user intent reduces reliability and can expose unrelated content if the translated keyword broadens the query.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
## 1. MCP Session and Authentication
- Use the already configured Gate MCP session for the current host.
- Local Gate MCP deployments use `GATE_API_KEY` and `GATE_API_SECRET`; never ask the user to paste these secrets into chat.
- Minimal permission for this skill is `Activity:Read`.
- If the required Gate MCP tools are missing, stop and switch to setup guidance only.
- If the MCP session returns an auth or permission error, stop and guide the user to repair the configured local MCP credentials before continuing.
The 'My Activities' flow instructs the agent to retrieve user-enrolled activities via cex_activity_get_my_activity_entry without any mention of authentication state, consent, or privacy disclosure. If invoked in the wrong context or surfaced to an unintended viewer, it could expose account-specific participation data and create a privacy leak.
The prompt example "Recommend some activities" is a very general phrase and the file does not define clear boundaries for when this skill should activate versus when a broader recommendation skill should handle the request. Because this is a markdown file, vague invocation language is in scope and may cause unintended activation on ordinary recommendation queries.
The example "What activities are available" is broad and could match many unrelated contexts, while the document does not provide negative examples or scope constraints. This makes it unclear when the skill should respond versus when another skill covering events, tasks, or general activities should be used.