Give teams shared coding-agent memory with Stash
Connect coding agents to a shared Stash knowledge base so sessions, files, pages, and team context can be searched, cited, updated, and reused across agent runs.
npx skills add agentskillexchange/skills --skill give-teams-shared-coding-agent-memory-with-stash
uv tool install stashai, then run stash signin in the repository you want to connect. For self-hosted use, deploy the documented Docker Compose stack, install the CLI, run stash signin, choose Self-host, and enter the self-hosted Stash URL.Use this workflow when a team needs coding agents to remember prior sessions, project decisions, files, and shared notes across machines and agent runtimes. Stash provides a CLI, MCP server, virtual filesystem shell, hooks, and optional self-hosting path so agents can ingest session transcripts, search pages and tables, retrieve cited answers, and write back useful context.
What this skill actually does
The operator should install the CLI, sign in or connect to a self-hosted instance, bind the current repository, and choose which supported coding agents get hooks during setup. After that, agents can run commands such as `stash vfs ls /`, `stash vfs “rg database migration /”`, or the documented CLI search and skills commands to recover previous work, answer questions with source-backed context, and share reusable skill bundles.
Invoke this instead of asking each agent to reconstruct context from git history, local notes, or one-off transcript files when the work spans long-running coding sessions, multiple teammates, or multiple agents such as Claude Code, Cursor, Codex, OpenCode, Gemini CLI, OpenClaw, or Hermes. Do not use it as a generic knowledge-base listing or as a replacement for source control, issue tracking, or permission review. The boundary is agent-operated shared memory: Stash stores and exposes context, while the agent still verifies facts against the repository and the operator controls privacy, transcript upload, integrations, and public skill sharing.