Memory & Context

Give agents a searchable memory of architecture, patterns and decisions

Design agent memory across context window, vector stores and long-term recall

Compact a long coding session at strategic points to save context

Search past coding agent sessions and check token use, tool breakdowns and tasks

Manage LLM context windows with summarizing, trimming and routing

Add short-term, long-term and entity memory to chat-based LLM apps

Keep versioned agent memory in a .GCC folder for multi-step, cross-session work

Search past conversations and semantic memory for earlier decisions

Cut LLM cost and latency with prompt caching and response caching

Process large logs, JSON and command output without filling the context window

Index local files or folders into a searchable knowledge base for later lookups

Permanently delete the context-mode knowledge base and reset session stats

Search indexed project docs and session memory for earlier notes and snippets

Show how much context window the session saved, with a per-tool breakdown

Add persistent memory and semantic search to AI apps with Supermemory

Check new knowledge pages for duplicates and entity conflicts before writing

Include working links to knowledge pages whenever you report or share them

Run lookups, enrichment and back-linking on a personal knowledge brain

Choose the correct folder and schema type for every new knowledge page

Sync ChatGPT and Claude conversation history into your knowledge brain on a schedule

Seed a new knowledge brain from email, calendar, contacts and social archives

Derive a sanitized shared team brain from a personal one by stripping private details

Merge concept stubs into a tiered map that traces how ideas evolved across sources

Audit always-loaded context files for redundancy, contradictions and wasted tokens

Import ChatGPT, Claude and Perplexity chat exports as dated archive pages

Trace each factual correction to its source and fix it so it cannot recur

Audit and file every output from a work session into the right knowledge pages

Create and update person and company pages with timelines and cross-links

Search and write a persistent company knowledge brain on people, projects and decisions

Get periodic advice on version drift, stalled jobs and setup gaps in your brain

Trace how one idea evolved, including reversals, contradictions and abandoned branches

Run brain health checks for stale info, orphan pages and broken back-links

Answer questions from your knowledge brain with search, synthesis and citations

Decide the right folder for each new knowledge file by its primary subject

Search the knowledge brain and the web before asking the user who someone is

Open the knowledge page for a mentioned entity before answering from memory

Add and refine page types in a knowledge brain's schema from real content

Migrate a knowledge brain to a canonical 14-type schema and merge noisy types

Opt in to passive capture of original ideas and entity mentions in messages

Save voice notes verbatim into the right knowledge folder without paraphrasing

Turn a finished voice call transcript into a brain page and post a summary

Apply saved coding preferences and record new ones from durable feedback

Look up and summarize local Deep Agents session threads and their titles

Save conventions, decisions and workflows from a session into AGENTS.md or skills

Install and verify the TencentDB memory plugin for OpenClaw long-term recall

Walk through connecting an AI agent to the Memory Proxy with checks at each step

Explain how agentmemory stores, retrieves and extends memory end to end

Explain and tune agentmemory hooks that capture observations automatically

Call the agentmemory HTTP REST API when MCP is unavailable or not supported