Documentation
¶
Overview ¶
Package tool contains domain types for tool discovery and risk classification.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type RiskLevel ¶
type RiskLevel string
RiskLevel represents the security risk level of a tool.
const ( // RiskLevelLow indicates read-only, informational operations. // Examples: list_files, get_status, help, version. RiskLevelLow RiskLevel = "LOW" // RiskLevelMedium indicates read operations with potential sensitivity. // Examples: fetch_data, download_file, export_report, search_users. RiskLevelMedium RiskLevel = "MEDIUM" // RiskLevelHigh indicates write operations or network access. // Examples: file_write, create_user, update_config, send_email. RiskLevelHigh RiskLevel = "HIGH" // RiskLevelCritical indicates destructive operations, system commands, or admin ops. // Examples: file_delete, execute_command, shell_exec, admin_reset. RiskLevelCritical RiskLevel = "CRITICAL" )
func ClassifyTool ¶
ClassifyTool determines the risk level of a tool based on its name. Classification is case-insensitive and uses pattern matching.
Priority order (highest to lowest):
- CRITICAL: destructive operations (delete, exec, shell, admin)
- HIGH: write operations (write, create, update, send)
- MEDIUM: sensitive reads (fetch, download, export, search)
- LOW: everything else (list, help, version)
Limitations:
- Uses simple substring matching (e.g., "undelete" also matches "delete")
- For v1, this is acceptable; admin overrides can address edge cases
- Tool descriptions are not analyzed, only names
type Tool ¶
type Tool struct {
// Name is the unique identifier for this tool (required).
Name string `json:"name"`
// Title is an optional human-readable display name.
Title *string `json:"title,omitempty"`
// Description is an optional human-readable description.
Description *string `json:"description,omitempty"`
// InputSchema is the JSON Schema for the tool's parameters (required).
InputSchema json.RawMessage `json:"inputSchema"`
// OutputSchema is an optional JSON Schema for the tool's output.
OutputSchema *json.RawMessage `json:"outputSchema,omitempty"`
// RiskLevel is the computed security risk level (not from MCP, added by classifier).
RiskLevel RiskLevel `json:"-"`
}
Tool represents a tool from the MCP tools/list response. Fields match the MCP specification 2025-06-18.
func ClassifyTools ¶
ClassifyTools returns a new slice of tools with RiskLevel populated on each. The input slice is not modified.
type ToolCatalog ¶
type ToolCatalog struct {
// Tools is the list of tools available from the server.
Tools []Tool `json:"tools"`
// NextCursor is the pagination cursor for fetching more tools.
NextCursor *string `json:"nextCursor,omitempty"`
// CachedAt is when this catalog was cached (UTC).
CachedAt time.Time `json:"cachedAt"`
// ServerID identifies which upstream server this catalog is from.
ServerID string `json:"serverId"`
}
ToolCatalog represents a cached collection of tools from an upstream MCP server.
Source Files
Click to show internal directories.
Click to hide internal directories.