These skills help an agent retain useful context between sessions and hand work to another agent. They cover Claude-Mem, Honcho, Memori, Obsidian-style recall, shared memory, and handoff notes.
28 skills
Pair the Claude-Mem Cowork plugin with a personal cmem.ai account and verify its masked configuration status.
Summarize a conversation into a handoff document for a fresh agent session.
Distill a conversation or completed task into reusable lessons for later agent work.
Save the decisions, facts, dead ends and architecture this conversation established to Atlas's shared memory, so other agents and future sessions can build on them.
Interview for a useful memory taxonomy, validate and install a custom mode, and verify it in startup context.
Create a new Claude Code session retaining the later half of an existing conversation.
Save player progress and leaderboards with guarded DataStore calls, UpdateAsync, retries, and shutdown handling.
Organize project rules, task context and references so coding agents receive relevant information at the right time.
Design serializable game state, save slots, crash-safe writes, schema upgrades, and throttled autosave.
Recall relevant context, summarize sessions and preserve durable memories through a configured Memori MCP connection.
Recall useful context and record durable insights with the connected Honcho memory integration.
Summarize a conversation and start a background Claude agent with the handoff.
Inspect and debug Honcho workspaces via the `honcho` CLI.
Write a precise HANDOFF.md covering the goal, current state, files, failed attempts, best theory, and next steps.
Write or update HANDOFF.md with goals, progress, successful attempts, dead ends and next steps.
Create a fresh Claude Code session retaining only the most recent quarter of a conversation.
Use Trail Code's graft index to inspect repository structure, relevant context and change impact.
Design agents around capabilities, domain knowledge, context and progressively added complexity.
Transfer a task to another agent with the working context and progress needed to continue.
Search previous Claude Code, Codex and Cursor sessions and summarize attempts, decisions and investigation history.
Integrate Honcho memory into existing Python or TypeScript codebases.
Install or remove memU through host-specific adapters that connect session recording and memory retrieval.
Persistent memory for coding agents, with 21 curated workflows for recall, code exploration, planning, reviews, reports, and integrations. Install the upstream plugin; hosted memory and model usage may cost extra.
Find previous decisions and solutions with a bounded index → context → selected-details search over Claude-Mem history.
Search Screenpipe OCR or audio history and optionally preserve selected findings in Claude-Mem.
Build a focused corpus from project observations, prime an AI session, and query that corpus conversationally.
Configure or diagnose Claude-Mem SyncHub through the official installer and verify authenticated hub reachability.
Set up Claude-Mem in Cursor with a local or remote worker and choose the appropriate observation provider.
Choose a handoff skill for a bounded transfer, a persistent store for long-lived context, and a search or recall skill when the information already exists somewhere.