SSH 동기화를 지원하는 코딩 에이전트용 오픈소스 메모리 계층

Your agents already solved this. deja finds it. vshulcz.github.io/deja-vu Claude Code, Codex and opencode write every conversation to local files — gigabytes of debugged problems and design decisions you can't search. deja is a zero-dependency binary that turns those histories into a memory layer: Search deja "connection pool exhausted" — 7–9 ms over gigabytes, retroactive: months of logs from before you installed it Agent recall MCP recall tool — the agent answers "we fixed this three weeks ago" instead of re-debugging, across harnesses Auto-recall install --auto adds a SessionStart hook: relevant memory lands in context before you ask Redaction API keys, JWTs, private keys are stripped at index time — the cache is safe to keep Stats deja stats — your agent work, wrapped: harnesses, top projects, activity sparkline Share deja share <id> — hand a colleague a sanitized digest of a session, secrets already scrubbed Sync deja sync export/import — move memory between machines, append-only, idempotent One binary. No models to download, no services to run, nothing leaves your machine. (opencode indexing shells out to the sqlite3 CLI, preinstalled on macOS and most Linux distros.) Install curl -fsSL https://raw.githubusercontent.com/vshulcz/deja-vu/main/install.sh | sh or: go install github.com/vshulcz/deja-vu/cmd/deja@latest # Go npx @vshulcz/deja-vu " query " # npm, no install brew install vshulcz/tap/deja-vu # Homebrew Wire it into the agents you use (edits config, keeps a .bak ): deja install --all # MCP recall for claude-code, codex, opencode deja install claude-code --auto # + SessionStart auto-recall hook That's it. Next session, ask your agent: have we dealt with jwt refresh rotation before? check your memory — or with --auto , don't ask: the agent starts each session already knowing what you solved in that project. CLI $ deja "jwt refresh token" [claude] api · Jul 8 · 8f31c0a9 — 2 matches login started failing after refresh token rotation; jwt kid mismatch in tests fixed by reloading jwks cache after rotateKey and adding a clock-skew test [codex] web · Jul 1 · b77d91e2 — 1 match refresh token cookie needed SameSite=Lax in local callback flow Command What it does deja <query> Search all histories. Multi-word = AND, substrings match ( code finds opencode ). --re , --harness , --project , --since 30d , --role , --json . deja ctx <query> Compact markdown digest of the best match — pipe it into a prompt. deja share <id> Sanitized session digest for a colleague: secrets redacted, tool noise stripped. deja stats Totals, per-harness split, top projects, monthly sparkline. --json too. deja sync export <dir> [--full] / import <dir> / ssh <host> [--pull] Move memory between machines — via a shared folder or one ssh command. Watermarked, append-only, idempotent. deja show <id> / deja last [n] Read one session / list recent ones. deja sources Discovered stores, sizes, message and redaction counts. deja mcp The stdio MCP server (what deja install wires in). deja warmup Build/refresh the index without searching — handy in cron or shell startup. Context piping without MCP: claude " Prior context: $( deja ctx ' database migration ' ) " Sync between machines Point both machines at one shared folder (Syncthing, iCloud, a git repo — anything that moves files): deja sync export ~ /Sync/deja # machine A: appends new batches since last export deja sync import ~ /Sync/deja # machine B: picks up what it hasn't seen Or skip the shared folder when the other machine is a ssh hop away: deja sync ssh mini # push new records to mini and import them there deja sync ssh mini --pull # fetch mini's new records into this machine ssh mode uses your system ssh/scp and the deja binary on the remote (looked up on PATH, falling back to ~/.local/bin/deja ). Batches are plain JSONL, redacted on the way out. Import is idempotent, so keep the folder as an append-only log and run both commands from cron if you like. Records never echo back to their origin. --full re-exports everything regardless of watermarks — useful when adding a machine after old batches are gone. Synced sessions show up under imported:<project> in search, recall , and session-start auto-recall. Teach your agent to remember deja install --all wires up MCP recall and the session-start hook. To make the agent reach for memory on its own, add this to your CLAUDE.md / AGENTS.md : Before debugging or re-implementing something, run `deja "<query>"` (or the MCP recall tool) — past agent sessions across Claude Code, Codex and opencode are indexed locally. Cite what you reuse. MCP tools Tool Arguments Returns recall query , harness? , limit? Dense matching snippets, ≤4KB — cheap on context. recall_context query Markdown digest of the best-matching session. With --auto , a SessionStart hook also feeds the current project's recent memory in automatically — read-only, capped at 2KB, and it never delays or breaks agent startup. Security Credentials are redacted at index time: AWS keys, generic api_key= / token= assignments, bearer tokens and raw JWTs, PEM private key blocks, provider tokens ( ghp_ , sk- , npm_ , xox. , AIza ), and scheme://user:pass@host URLs. The value is replaced with [redacted:<kind>] ; surrounding text stays searchable. deja sources shows per-store counts. Opt out with DEJA_NO_REDACT=1 (unsafe). deja share and deja sync export re-apply redaction on the way out. Supported harnesses Harness Store Status Claude Code ~/.claude/projects/**/*.jsonl ✅ Codex CLI ~/.codex/sessions/** + history.jsonl ✅ opencode ~/.local/share/opencode/opencode.db ✅ aider, Gemini CLI #6 , #7 planned Custom locations via DEJA_CLAUDE_ROOT , DEJA_CODEX_ROOT , DEJA_OPENCODE_DB , DEJA_INDEX_DIR . Performance Measured on a real corpus — 1,250+ sessions, ~3.3GB across three harnesses: Warm search 7–9 ms typical, ~40 ms worst-case Cold index (once) ~10 s Index size ~2.4% of corpus The index is incremental: when a session file grows, only that file is re-read. How it works Local inverted index in ~/.cache/deja : parse JSONL/SQLite stores → redact credentials → records.bin + token buckets → manifest.json tracks per-file state so repeat runs only ingest what changed. The MCP server, stats, share and sync all read the same index. Details: docs/ARCHITECTURE.md . Privacy: no network path exists in the indexing or search code. Local files in, local cache out. FAQ Does anything leave my machine? No. There is no network code in the tool. sync writes files to a directory you choose; moving them is up to you. How is this different from cass? cass is the kitchen-sink take on session search: 22 providers, Rust, optional semantic embeddings, a TUI. deja is the opposite bet — one small Go binary, pure lexical, three harnesses, zero setup — plus the memory-layer pieces around it: auto-recall, redaction, share, sync. And from MemPalace / Mem0 / Letta? Those are memory platforms : embeddings, vector stores, capture hooks or APIs that record going forward. deja has no capture step at all — it indexes what your agents already wrote to disk, including months of history from before you installed it. They can coexist. What about secrets already in my logs? They stay in the original harness files (that's your agent's data), but they don't enter deja's index, digests, shares or sync exports. What about Windows? Builds exist, CI runs the suite on Windows; macOS/Linux are the battle-tested paths. Field reports welcome: #9 . Can I exclude a project? Not yet — planned as --exclude ( #8 ). Today you can point DEJA_*_ROOT at a filtered copy. How do I wipe everything? deja uninstall --all rm -rf ~ /.cache/deja Contributing make build test lint — see CONTRIBUTING.md . Adding a harness is one parser file: docs/ARCHITECTURE.md . Good first issues are labeled. License MIT © Vladislav Shulcz