Memory & Context

Search your MemPalace before answering questions about past work or decisions

Build self-improving agent loops that learn from past run traces safely

Inject a status bar into long agent runs to stop loops and goal drift

Compress long agent context using six strategies and layered compaction

Structure agent knowledge with GraphRAG, RAPTOR and layered memory indexes

Keep KV cache prefixes stable to cut agent latency and token cost

Design user memory for agents, from plain notes to structured memory cards

Hand off work between bots with a durable shared record in Hindsight memory

Plan where memory adds value in your app, with bank config, tag schema and code

Store team conventions and lessons in Hindsight Cloud and recall them before tasks

Configure Hindsight coding memory per repo and troubleshoot memory issues

Read Hindsight docs on architecture, APIs, configuration and recommended patterns

Store user preferences and task outcomes in local Hindsight memory

Recall past context and retain durable facts across sessions with Hindsight memory

Search long-term memory from past coding sessions with Hindsight MCP tools

Store team conventions and learnings in a self-hosted Hindsight memory server

Load relevant long-term memory from Hindsight before starting any task

Answer questions about past decisions and habits by reasoning over Hindsight memory

Save facts, decisions and lessons to Hindsight memory as tasks finish

Set up Hindsight memory for this bot, check it and fix failing memory calls

Carry state between scheduled routine runs using Hindsight memory

Search past conversations in the history log to recall earlier context

Keep a Markdown wiki where every claim traces back to an immutable source

Keep long agent sessions on track and prevent rule drift after compaction

Keep project state and confirmed decisions in Superself for agent sessions

Load relevant experience memories from OpenViking before starting a task

Install the OpenViking memory plugin for OpenClaw in a guided chat setup

Work with OpenViking from OpenClaw for memory, session archives and skill import

Recall past sessions, save facts and files, and forget items in OpenViking memory

Route the ov dream and ov recall commands to the OpenViking sync and recall CLI

Retrieve and apply past experience memories before multi-step coding or config work

Trace wrong, missing or changed OpenViking memories back to their source archive

Add, browse, search, back up and remove resources stored in OpenViking

Map every file relevant to a task before changing anything

Load project context from Copilot Spaces to ground answers in shared docs

Write and read journal entries that carry work state across agent sessions

Create a new desk folder with an initial journal so later sessions can resume

Fold mature lessons from a memory file into the matching instructions file

Build a persistent knowledge base that links ingested documents into a graph

Save lessons learned as domain-organized memory instructions for future sessions

Build a personal context portfolio through an interview so agents know how you work

Compress responses with shorthand to cut output tokens while keeping technical detail

Have Copilot list the files it needs to see before answering your question

Keep an organizational context wiki coherent by working through its lint report queue

Synthesize recent team activity into a durable, sourced context wiki on a schedule

Triage context wiki quality issues like contradictions, stale priorities and duplicates

Sort reusable project knowledge into memory, notes or durable docs

Map the current session's goal, pivots and open decisions to get back on track

Build a persistent markdown wiki that compounds knowledge across sessions