[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:ai-memory":3},"\u003Cp align=\"center\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"docs\u002Flogo-dark.png\">\u003C\u002Fsource>\n    \u003Cimg alt=\"ai-memory\" src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fakitaonrails\u002Fai-memory\u002FHEAD\u002Fdocs\u002Flogo-light.png\" width=\"480\" \u002F>\n  \u003C\u002Fpicture>\n\u003C\u002Fp>\u003Cblockquote>\n\u003Cp>Long-term memory for AI coding agents. Quit Claude Code mid-task,\nstart OpenAI Codex in the same directory, continue without\nre-explaining the architecture, the failed approaches, or the open\nquestions.\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fakitaonrails\u002Fai-memory\u002Freleases\u002Flatest\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fakitaonrails\u002Fai-memory\" alt=\"Release\" \u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fakitaonrails\u002Fai-memory\u002Fblob\u002FHEAD\u002Frust-toolchain.toml\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Frust-1.95+-blue\" alt=\"Rust\" \u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fakitaonrails\u002Fai-memory\u002Fblob\u002FHEAD\u002FLICENSE\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-blue\" alt=\"License\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Ch2>Support Matrix\u003C\u002Fh2>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Area\u003C\u002Fth>\n\u003Cth>Status\u003C\u002Fth>\n\u003Cth>Notes\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Linux\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>Primary Docker\u002Fserver target and CI platform. Published Docker images support \u003Ccode>linux\u002Famd64\u003C\u002Fcode> and \u003Ccode>linux\u002Farm64\u003C\u002Fcode>. Native Arch\u002FAUR packages include system and user systemd units.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>macOS\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>Workspace tests run in CI; tagged releases publish native \u003Ccode>ai-memory-macos-aarch64.tar.gz\u003C\u002Fcode> and \u003Ccode>ai-memory-macos-x86_64.tar.gz\u003C\u002Fcode> binaries. The native binary is the recommended path on Apple Silicon. See \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fakitaonrails\u002Fai-memory\u002Fblob\u002FHEAD\u002Fdocs\u002Fmacos.md\" rel=\"nofollow ugc noopener\">\u003Ccode>docs\u002Fmacos.md\u003C\u002Fcode>\u003C\u002Fa>.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Windows via WSL2\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>Use the Linux install path inside WSL2 when the agent runs there.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Native Windows\u003C\u002Ftd>\n\u003Ctd>Experimental\u003C\u002Ftd>\n\u003Ctd>Tagged releases publish \u003Ccode>ai-memory-windows-x86_64.zip\u003C\u002Fcode> with \u003Ccode>ai-memory.exe\u003C\u002Fcode>; Docker Desktop wrapper and source builds are also available. Local supported profiles default to host-native hook commands; Claude Code may use its Windows exec form, while other agents use native single command strings matching their hook schema. PowerShell\u002FGit Bash scripts are compatibility fallbacks. See \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fakitaonrails\u002Fai-memory\u002Fblob\u002FHEAD\u002Fdocs\u002Fwindows.md\" rel=\"nofollow ugc noopener\">\u003Ccode>docs\u002Fwindows.md\u003C\u002Fcode>\u003C\u002Fa>.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Claude Code\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>MCP config + lifecycle hooks; native commands enforce capture exclusions. \u003Ccode>install-mcp --session-aware\u003C\u002Fcode> optionally enables per-session auto-scope isolation through a local stdio bridge. Optionally captures the assistant's final turn on \u003Ccode>Stop\u003C\u002Fcode> when installed with \u003Ccode>--capture-assistant\u003C\u002Fcode> and the server enables \u003Ccode>capture_assistant\u003C\u002Fcode> (double opt-in, off by default).\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Codex\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>MCP config + lifecycle hooks; native commands enforce capture exclusions. No automatic true session-end hook, so run \u003Ccode>ai-memory finalize-session\u003C\u002Fcode> when you need a final summary\u002Fhandoff.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Command Code\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>MCP config (\u003Ccode>~\u002F.commandcode\u002Fmcp.json\u003C\u002Fcode>) + its four stable lifecycle-hook events (\u003Ccode>~\u002F.commandcode\u002Fsettings.json\u003C\u002Fcode>); native commands enforce capture exclusions and \u003Ccode>SessionStart\u003C\u002Fcode> injects handoffs. \u003Ccode>Stop\u003C\u002Fcode> is only a turn boundary, so use \u003Ccode>ai-memory finalize-session --agent command-code\u003C\u002Fcode> after the final turn. \u003Ccode>ai-memory run command-code\u003C\u002Fcode> adds exact v3 native-session resume and visible-event import; experimental unsandboxed Mods remain excluded.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Devin CLI\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>MCP config + lifecycle hooks. Hooks use Devin's \u003Ccode>PostCompaction\u003C\u002Fcode> event, inject handoffs via \u003Ccode>hookSpecificOutput.additionalContext\u003C\u002Fcode>, and omit subagent events because Devin does not expose them.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>OpenCode\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>Remote MCP config + generated TypeScript plugin; generated plugin enforces capture exclusions.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Cursor\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>MCP config + lifecycle hooks.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Gemini CLI\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>MCP config + lifecycle hooks.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Oh My Pi \u002F OMP\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>Use \u003Ccode>--client omp\u003C\u002Fcode> \u002F \u003Ccode>--agent omp\u003C\u002Fcode> (or \u003Ccode>oh-my-pi\u003C\u002Fcode>) for native \u003Ccode>.omp\u003C\u002Fcode> MCP config + TypeScript extension; generated extension enforces capture exclusions.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Pi\u003C\u002Ftd>\n\u003Ctd>Supported\u003C\u002Ftd>\n\u003Ctd>Generated \u003Ccode>~\u002F.pi\u002Fagent\u002Fextensions\u002Fai-memory-pi.ts\u003C\u002Fcode> extension provides lifecycle capture and an HTTP MCP bridge; generated extension enforces capture exclusions.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Crush\u003C\u002Ftd>\n\u003Ctd>Managed-only\u003C\u002Ftd>\n\u003Ctd>\u003Ccode>ai-memory run crush\u003C\u002Fcode> resumes its project-local session database and supplies portable context through a temporary supported global-context file; no lifecycle-hook installer is provided.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Managed workstreams\u003C\u002Ftd>\n\u003Ctd>Opt-in\u003C\u002Ftd>\n\u003Ctd>\u003Ccode>ai-memory run\u003C\u002Fcode> provides transparent cross-harness continuity for Claude Code, Codex, Open\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\n",1787581364406]