Memory & Context

Diagnose and configure MemSearch memory, index health and provider settings

Recall past decisions, debugging notes and conversations from memory search

Audit token waste and dollar cost across several coding agent systems

Restore a prior session checkpoint as scrubbed context when resuming work

Coach a leaner coding agent setup from your own usage data to cut token costs

Price what a past agent session cost and open the token usage dashboard

Find context-window waste in an agent setup, fix it with approval and measure savings

Filter private command output before it enters the agent context to save tokens

Connect OpenViking memory with browser automation and web search tools

Log errors, user corrections and missing capabilities so future sessions improve

Extract successes, failures and false assumptions from agent runs into learnings

Build and maintain a personal LLM wiki from raw sources with citations and linting

Write handoff documents that carry context into the next agent session

Surface open questions and dream-synthesized insights from past explorations

Explain how memory, dream consolidation and knowledge crystals work

Force a memory search before stating facts that have not been checked yet

Route relationship questions about people to the knowledge graph, not search

Decide what to save to working memory and when to record corrections

Archive every file an agent writes with session IDs, timestamps and read-back checks

Search and feed a local knowledge base with document retrieval and mode switching

Store structured agent memory as a typed knowledge graph of entities and links

Query your own session history, token usage and cost data from the session database

Make an agent review its own mistakes and keep lessons in its memory

Route stable preferences and decisions into AGENTS.md files and memory notes

Convert long session logs into readable Markdown and read them in chunks

Design agent context, memory and token budgets for multi-agent systems and LLM pipelines

Pack a whole code repository into one AI-friendly XML, Markdown or text file

Compile documents and notes into a citation-traceable wiki for agent retrieval

Administer Supermemory spaces and memory entries from the command line

Turn a recurring workflow or messy folder into an agent-runnable workspace

Recover the full context of the last task from a parent session transcript

Search earlier coding agent sessions for past work, decisions and code patterns

Save user preferences, successful patterns and error lessons to long-term memory

Read recent chat messages, replies and task notifications from the current chat

Check how many memories are stored and how they break down by type

Read the saved user profile with preferences and context for personal replies

Search saved memories by keyword and type to recall preferences and lessons

Stop asking about a profile detail when the user declines to answer it

Save user preferences, habits and work details to a personal profile

Turn session observations into confidence-scored instincts that become skills

Refine context retrieval step by step so subagents get the right information

Record the decisions made in a session into the wiki pages they affect

Pick which team or local wiki a session uses and link a repo to its project

Answer what was decided and why from the project wiki before changing an area

Create a project wiki with a home page, starter pages and a writing guide

Learn how agents build a local wiki, capture decisions and open the browser view

Move a project wiki onto Nimbalyst Teams and invite teammates to view it

Write and maintain wiki pages, decisions and links alongside code changes

Capture corrections and preferences during sessions and save them to CLAUDE.md