Documentation
¶
Overview ¶
Package models defines data structures for the storage system.
Package models defines data structures for the storage system.
Package models defines data structures for the storage system. This package contains domain models for knowledge chunks, experiences, tools, conversations, task results, and secrets.
Package models defines data structures for the storage system.
Package models defines data structures for the storage system.
Package models defines data structures for the storage system.
Index ¶
Constants ¶
const ( RoleSystem = "system" RoleUser = "user" RoleAssistant = "assistant" RoleTool = "tool" )
ConversationRole constants.
const ( // ExperienceTypeSuccess represents a successful experience. ExperienceTypeSuccess = "success" // ExperienceTypeFailure represents a failed experience. ExperienceTypeFailure = "failure" // ExperienceTypeQuery represents a query-based experience (for backward compatibility). ExperienceTypeQuery = "query" // ExperienceTypeSolution represents a solution-based experience (for backward compatibility). ExperienceTypeSolution = "solution" // ExperienceTypePattern represents a pattern-based experience (for backward compatibility). ExperienceTypePattern = "pattern" // ExperienceTypeDistilled represents a distilled experience (for backward compatibility). ExperienceTypeDistilled = "distilled" )
ExperienceType constants.
const ( EmbeddingStatusPending = "pending" EmbeddingStatusProcessing = "processing" EmbeddingStatusCompleted = "completed" EmbeddingStatusFailed = "failed" )
EmbeddingStatus constants.
const ( TaskStatusPending = "pending" TaskStatusRunning = "running" TaskStatusCompleted = "completed" TaskStatusFailed = "failed" )
TaskStatus constants.
const (
AlgorithmAESGCM = "aes-gcm"
)
EncryptionAlgorithm constants.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Conversation ¶
type Conversation struct {
ID string `json:"id"`
SessionID string `json:"session_id"`
TenantID string `json:"tenant_id"`
UserID string `json:"user_id"`
AgentID string `json:"agent_id"`
Role string `json:"role"`
Content string `json:"content"`
ExpiresAt time.Time `json:"expires_at"`
CreatedAt time.Time `json:"created_at"`
}
Conversation represents a chat message in a conversation session. This stores short-term conversation context without vector embedding.
func (*Conversation) IsExpired ¶
func (c *Conversation) IsExpired() bool
IsExpired checks if the conversation has expired based on TTL.
func (*Conversation) TableName ¶
func (c *Conversation) TableName() string
TableName returns the table name for this model.
type Experience ¶
type Experience struct {
// ID is the unique identifier of the experience.
ID string `json:"id"`
// TenantID is the tenant identifier for multi-tenancy isolation.
TenantID string `json:"tenant_id"`
// Type is the type of experience.
// Valid values: "success" or "failure".
Type string `json:"type"`
// Problem is the abstract problem statement.
// This is the target for embedding generation.
// NOTE: This field is stored in the 'input' column for backward compatibility.
Problem string `json:"problem"`
// Solution is the concise solution approach.
// NOTE: This field is stored in the 'output' column for backward compatibility.
Solution string `json:"solution"`
// Input is the raw input text (stored in database 'input' column).
// This is for backward compatibility with the database schema.
Input string `json:"-"`
// Output is the raw output text (stored in database 'output' column).
// This is for backward compatibility with the database schema.
Output string `json:"-"`
// Constraints are important constraints or context for the solution.
// NOTE: This field is stored in metadata['constraints'] for backward compatibility.
Constraints string `json:"constraints"`
// Embedding is the vector embedding of the problem.
// This is used for semantic similarity search.
Embedding []float64 `json:"embedding"`
// EmbeddingModel is the name of the embedding model used.
EmbeddingModel string `json:"embedding_model"`
// EmbeddingVersion is the version of the embedding model.
EmbeddingVersion int `json:"embedding_version"`
// Score is the overall score of the experience.
Score float64 `json:"score"`
// Success indicates whether the original task was successful.
Success bool `json:"success"`
// AgentID is the identifier of the agent that generated this experience.
AgentID string `json:"agent_id"`
// UsageCount is the number of times this experience was successfully used.
// This is used as a reinforcement signal.
// NOTE: This field is stored in metadata['usage_count'] for backward compatibility.
UsageCount int `json:"usage_count"`
// Metadata is additional metadata for the experience.
// NOTE: For backward compatibility, constraints and usage_count are also stored here.
Metadata map[string]interface{} `json:"metadata"`
// DecayAt is the time when this experience should be considered expired.
// If zero, the experience never expires.
DecayAt time.Time `json:"decay_at"`
// CreatedAt is the time when this experience was created.
CreatedAt time.Time `json:"created_at"`
// UpdatedAt is the time when this experience was last updated.
UpdatedAt time.Time `json:"updated_at"`
}
Experience represents a distilled task experience. This is the core data structure for experience storage and retrieval. The structure follows the principle: Experience = Distilled Knowledge.
func (*Experience) GetConstraints ¶
func (e *Experience) GetConstraints() string
GetConstraints retrieves constraints from metadata. This provides backward compatibility for the new structure.
func (*Experience) GetUsageCount ¶
func (e *Experience) GetUsageCount() int
GetUsageCount retrieves usage count from metadata. This provides backward compatibility for the new structure.
func (*Experience) IsExpired ¶
func (e *Experience) IsExpired() bool
IsExpired checks if the experience has decayed and should be excluded from search.
func (*Experience) TableName ¶
func (e *Experience) TableName() string
TableName returns the table name for this model.
type KnowledgeChunk ¶
type KnowledgeChunk struct {
ID string `json:"id"`
TenantID string `json:"tenant_id"`
Content string `json:"content"`
Embedding []float64 `json:"embedding"`
EmbeddingModel string `json:"embedding_model"`
EmbeddingVersion int `json:"embedding_version"`
EmbeddingStatus string `json:"embedding_status"`
ChunkIndex int `json:"chunk_index"`
DocumentID string `json:"document_id"`
SourceType string `json:"source_type"`
Source string `json:"source"`
Metadata map[string]interface{} `json:"metadata"`
ContentHash string `json:"content_hash"`
AccessCount int `json:"access_count"`
CreatedAt time.Time `json:"created_at"`
UpdatedAt time.Time `json:"updated_at"`
}
KnowledgeChunk represents a RAG knowledge base entry with vector embedding. This is the core data structure for semantic search and retrieval.
func (*KnowledgeChunk) TableName ¶
func (k *KnowledgeChunk) TableName() string
TableName returns the table name for this model. Different dimensions use different tables to avoid mixing vector spaces.
type Secret ¶
type Secret struct {
ID string `json:"id"`
TenantID string `json:"tenant_id"`
Key string `json:"key"`
Value []byte `json:"value"`
KeyVersion int `json:"key_version"`
Algorithm string `json:"algorithm"`
ExpiresAt time.Time `json:"expires_at"`
Metadata map[string]interface{} `json:"metadata"`
CreatedAt time.Time `json:"created_at"`
}
Secret represents sensitive data with encryption. This stores API keys, passwords, and other sensitive information with AES-GCM encryption.
type TaskResult ¶
type TaskResult struct {
ID string `json:"id"`
TenantID string `json:"tenant_id"`
SessionID string `json:"session_id"`
TaskType string `json:"task_type"`
AgentID string `json:"agent_id"`
Input map[string]interface{} `json:"input"`
Output map[string]interface{} `json:"output"`
Embedding []float64 `json:"embedding"`
EmbeddingModel string `json:"embedding_model"`
EmbeddingVersion int `json:"embedding_version"`
Status string `json:"status"`
Error string `json:"error"`
LatencyMs int `json:"latency_ms"`
Metadata map[string]interface{} `json:"metadata"`
CreatedAt time.Time `json:"created_at"`
}
TaskResult represents the execution result of an agent task. This stores task outputs with vector embedding for future reference and learning.
func (*TaskResult) IsFailed ¶
func (t *TaskResult) IsFailed() bool
IsFailed checks if the task execution failed.
func (*TaskResult) IsSuccessful ¶
func (t *TaskResult) IsSuccessful() bool
IsSuccessful checks if the task execution was successful.
func (*TaskResult) TableName ¶
func (t *TaskResult) TableName() string
TableName returns the table name for this model.
type Tool ¶
type Tool struct {
ID string `json:"id"`
TenantID string `json:"tenant_id"`
Name string `json:"name"`
Description string `json:"description"`
Embedding []float64 `json:"embedding"`
EmbeddingModel string `json:"embedding_model"`
EmbeddingVersion int `json:"embedding_version"`
AgentType string `json:"agent_type"`
Tags []string `json:"tags"`
UsageCount int `json:"usage_count"`
SuccessRate float64 `json:"success_rate"`
LastUsedAt time.Time `json:"last_used_at"`
Metadata map[string]interface{} `json:"metadata"`
CreatedAt time.Time `json:"created_at"`
}
Tool represents an agent tool with semantic embedding for intelligent selection. Tools are stored with embedding to enable semantic matching beyond keyword search.
func (*Tool) IsAvailable ¶
IsAvailable checks if the tool is available for use based on success rate.
func (*Tool) UpdateUsage ¶
UpdateUsage updates the usage statistics after tool execution.
Source Files