Package com.tnsai.planning.tasklist

Interface ProgressMemory

public interface ProgressMemory

Persistent learning store for cross-task knowledge transfer.

ProgressMemory implements the Ralph Pattern's "Progress Memory" principle: learnings from completed tasks are preserved and injected as context into subsequent tasks, enabling iterative improvement across a task list.

Usage

 ProgressMemory memory = ProgressMemory.inMemory();

 // Record a learning
 memory.recordLearning("task-1", "JWT requires RS256 for production",
     Set.of("auth", "security"));

 // Get relevant learnings for next task
 List<Learning> relevant = memory.getRelevantLearnings(Set.of("auth"));

 // Build context prompt for LLM
 String context = memory.buildContextPrompt();
 

Author:

TnsAI Team

See Also:

• TaskListExecutor

Nested Class Summary

Nested Classes

Modifier and Type	Interface	Description
static final record	ProgressMemory.Learning	A recorded learning from task execution.

Method Summary

Modifier and Type	Method	Description
String	buildContextPrompt()	Builds a context prompt from all stored learnings.
void	clear()	Clears all stored learnings.
static ProgressMemory	file(Path basePath)	Creates a file-based progress memory.
List<ProgressMemory.Learning>	getLearnings()	Gets all recorded learnings.
List<ProgressMemory.Learning>	getRelevantLearnings(Set<String> tags)	Gets learnings relevant to the specified tags.
static ProgressMemory	inMemory()	Creates an in-memory progress memory.
void	recordLearning(String taskId, String content, Set<String> tags)	Records a learning from a completed task.
int	size()	Returns the number of stored learnings.

Method Details

inMemory

static ProgressMemory inMemory()

Creates an in-memory progress memory.

Returns:

new in-memory progress memory

file

static ProgressMemory file(Path basePath)

Creates a file-based progress memory.

Parameters:

basePath - directory to store progress data

Returns:

new file-based progress memory

recordLearning

void recordLearning(String taskId,
 String content,
 Set<String> tags)

Records a learning from a completed task.

Parameters:

taskId - the task that produced this learning

content - the learning content

tags - tags for relevance matching

getLearnings

List<ProgressMemory.Learning> getLearnings()

Gets all recorded learnings.

Returns:

all learnings, ordered by recording time

getRelevantLearnings

List<ProgressMemory.Learning> getRelevantLearnings(Set<String> tags)

Gets learnings relevant to the specified tags.

A learning is relevant if it shares at least one tag with the query.

Parameters:

tags - the tags to match

Returns:

matching learnings

buildContextPrompt

String buildContextPrompt()

Builds a context prompt from all stored learnings.

The returned string is formatted for injection into an LLM prompt, summarizing prior learnings for context transfer.

Returns:

formatted context string, or empty string if no learnings

size

int size()

Returns the number of stored learnings.

clear

void clear()

Clears all stored learnings.