Operations

Runtime Assumptions#

The normal non-AI runtime should work without a model provider. Daemon startup, CLI commands, MCP startup, sessions, docs, context preview, import, inbox, backup, trash, search, and the saved Graph context map do not require LM Studio, Ollama, llama.cpp, or a remote provider.

The uncommitted 2026-08-12 working tree passed the workspace typecheck, 333 automated tests with zero failures and two intentional Windows symlink-safety skips, the production web build and unchanged bundle budgets, isolated secret-canary, fixture, source-artifact, accessibility-source, and public-doc checks, plus a headless Edge smoke confirming the UI reports a missing daemon accurately at startup. Those local results are not commit-bound release evidence. Six Rust unit tests and the Tauri compile/package mechanics passed with an inert sidecar fixture, but the production daemon/runtime, live-provider, assistive-device, installer/signing, and rollback gates also remain open for an exact candidate.

The Vite production build uses native Rollup/esbuild optional packages. A checkout shared between Windows and WSL must install dependencies in the same operating system that will run the build; otherwise Vite can fail with a missing or wrong-platform native package.

Native Tauri/Rust validation depends on a local Rust toolchain. A candidate release executable is written to apps/desktop/src-tauri/target/release/zharwing-memory-desktop.exe; its build, installer, signing, and device smoke must be proven for that exact candidate.

Environment Variables#

Install dependencies, then create a private, untracked .env and choose a private local memory root. Do not copy retired browser settings into a new configuration:

corepack pnpm install

Daemon:

ZHARWING_MEMORY_HOST=127.0.0.1
ZHARWING_MEMORY_PORT=37841
ZHARWING_MEMORY_ROOT=<memory-root>
ZHARWING_MEMORY_AGENT_SURFACE=enabled
ZHARWING_MEMORY_DESKTOP_AUTOSTART_DAEMON=true

ZHARWING_MEMORY_ROOT stores private project memory, sessions, docs, imports, context bundles, Memory Inbox proposals, and backups. Do not commit this folder.

API client:

ZHARWING_MEMORY_DAEMON_URL=http://127.0.0.1:37841

Normal personal-preview mode needs no token. When advanced token mode or hardened-local is selected and ZHARWING_MEMORY_AUTH_TOKEN is unset, the daemon generates a random per-user token and stores it in the OS user state directory (%APPDATA%\zharwing-memory\daemon-token on Windows, $XDG_STATE_HOME or ~/.local/state/zharwing-memory/daemon-token on POSIX). Delete the file to rotate the token. Never commit a real token or use a placeholder token outside local development.

Optional desktop-shell overrides use ZHARWING_MEMORY_DESKTOP_PROJECT_ROOT, ZHARWING_MEMORY_DESKTOP_DAEMON_COMMAND, and ZHARWING_MEMORY_DESKTOP_AUTOSTART_DAEMON. Canonical ZHARWING_MEMORY_* variables always win. The previous AIMEM_* names remain fallback-only for one compatibility window.

Rename Compatibility#

New configuration, generated MCP files, logs, schemas, skill metadata, and runtime metadata use the Zharwing Memory name. A small set of legacy identifiers is retained deliberately so existing installations do not lose data or become separate applications:

• AIMEM_* environment variables are read only as deprecated fallbacks.
• aimem remains a CLI alias during the transition.
• .ai-memory.json remains a readable legacy pointer filename.
• AI Memory Root, local.aimem.desktop, and existing aimem.* browser storage keys remain stable compatibility identifiers.

Do not use those legacy identifiers in new examples or generated configuration.

Intended Startup#

Daemon:

corepack pnpm dev:daemon

CLI:

corepack pnpm dev:cli projects

Local Browser UI#

The browser UI is a full local interface for daily use. Start the daemon and UI together:

corepack pnpm dev

Open http://127.0.0.1:5174/. The React/Vite dev server is pinned to that port so it does not collide with other local product runtimes that commonly use Vite's default 5173.

Normal local use requires no token, launcher, or profile setup. See Browser UI.

Desktop/web UI workflows currently include setup, project selection, project delete, repo links, import preview/commit, workstreams, sessions, docs, search, context preview, inbox, graph, backup management, and Trash restore/purge. The browser UI calls the daemon API, so the daemon must be running first in browser mode. The native Tauri host refuses an unrelated healthy daemon. It starts and owns an exact-loopback hardened daemon, exchanges a one-shot desktop credential outside the webview, and rotates the daemon/principal when project binding changes. A packaged executable needs its sidecar or an explicit trusted ZHARWING_MEMORY_DESKTOP_DAEMON_COMMAND.

When opened as a native Tauri window, the Setup, Repositories, and Import screens use OS folder pickers for path fields. Browser dev mode leaves those Browse buttons disabled and keeps typed paths as the fallback, because ordinary browsers cannot expose arbitrary absolute folder paths to web apps.

Use corepack pnpm dev for the browser app. Use corepack pnpm dev:desktop for the native Tauri window. The Tauri dev command starts or reuses the Vite app on port 5174; the Rust desktop host owns its daemon on 127.0.0.1:37841.

Native Tauri dev window:

corepack pnpm dev:desktop

Packaged desktop build:

corepack pnpm build:desktop

On Windows this produces apps/desktop/src-tauri/target/release/zharwing-memory-desktop.exe. The current repository config does not build an installer.

MCP:

corepack pnpm dev:mcp

Installed client setup:

zharwing-memory mcp install auto
zharwing-memory mcp doctor

Use HTTP MCP when the client can reach the daemon at http://127.0.0.1:37841/mcp. Use stdio with --transport stdio when a client needs to launch a local subprocess or when Windows/WSL localhost routing makes the daemon URL unreachable. See MCP Setup.

The daemon and browser UI commands are source entry points, not validation evidence. The native Tauri command requires the local Rust/Tauri toolchain.

Security#

Current daemon behavior:

• generates a per-user local auth token only when advanced token mode is active and no token was supplied, storing it with restrictive permissions
• binds to localhost by default

Production packaging should additionally:

• allow ZHARWING_MEMORY_AUTH_MODE=none only for loopback-only personal setups
• keep remote access disabled by default
• require confirmation for project creation, repo linking, and destructive administrative operations
• serve selected-project memory to AI by default without per-request approval; keep explicit visibility exclusions, never-send patterns, and secret checks
• route canonical memory writes through review only when the project enables review mode or the update is risky or uncertain
• log sensitive operations without storing raw secrets

Backups And Trash#

Current backup behavior creates a directory snapshot under:

<project-memory-root>/backups/snapshots/<timestamp>/

The backup function excludes the backups/ subtree to avoid recursively copying snapshots into snapshots.

Future archive export can add .zip packaging once dependency or platform APIs are available.

Deleting a backup snapshot moves it to global Trash first. Trash metadata is stored under:

<memory-root>/global/trash/items/<trash-id>/trash-item.json

Path-backed items are moved below the same trash item directory. JSON-backed items, such as linked repo entries, store their payload in the trash item directory. Trash items can be restored until permanently purged.

Indexing#

Current index behavior writes a versioned project manifest:

generated/index.json

The index includes its schema version, generation time, collection counts, and compact record projections. It is rebuildable from Markdown/frontmatter and JSON proposal files and is the supported default for normal project sizes.

Optional scaling work:

• add SQLite registry/index database
• add SQLite FTS5 keyword search
• extend the existing rebuild-index command with a richer validation report

Local Assistant#

Current assistant behavior:

• status reporting
• recommended model metadata
• automatic close-session TLDR generation
• manual one-session TLDR generation
• bulk missing/all session TLDR generation
• deterministic return summary proposal
• deterministic document classification proposal

Deliberately out of scope:

• downloading or launching llama.cpp
• downloading GGUF models
• managing GPU acceleration
• prompt logs
• remote session TLDR approval flow

The UI supports configured LM Studio, Ollama, llama.cpp server, OpenAI, Anthropic, and custom OpenAI-compatible endpoints. It does not advertise an app-managed download/runtime path. Persisted legacy app-managed-llamacpp settings remain readable and show a migration message.

AI Provider Smoke Testing#

Start LM Studio or another local OpenAI-compatible provider when testing model-backed session TLDR generation, semantic graph relationship review, or provider connectivity. It is not needed for normal validation or Graph context map viewing.

Use Testing With AI Providers for the full runbook. The short version is:

1. Start the daemon.
2. Start the provider and load a JSON-capable local model.
3. Configure the endpoint and model in Settings -> Assistant.
4. Run Test provider or memory.check_semantic_graph_provider.
5. Close a session or use Work -> Sessions -> Generate TLDR to test session summarization.
6. Open Graph -> Details and click Run review.
7. Inspect the Inbox proposal before accepting edges.

Dry-run mode, provider overrides, and document/candidate limits are available under Graph Details Advanced.

Validation Checklist For Future Runtime Work#

1. Install dependencies.
2. Run corepack pnpm check:source-artifacts.
3. Run corepack pnpm typecheck.
4. Run corepack pnpm test (compiles from clean and fails on zero tests).
5. Run corepack pnpm build after dependencies are installed for the current OS.
6. Run corepack pnpm check:source-artifacts again to confirm no build wrote output into a src directory.
7. Start daemon.
8. Initialize a temporary project.
9. Start a session.
10. Generate context preview.
11. Save checkpoint.
12. Close session.
13. Create inbox proposal.
14. Rebuild index.
15. Create backup snapshot.
16. Open desktop app.
17. Connect MCP adapter to a client.
18. If testing AI features, run the provider smoke checklist in Testing With AI Providers.

Known Constraints#

• The JSON project manifest is the supported default. SQLite/FTS5 remains an optional scaling layer for very large stores.
• MCP adapter is dependency-free and does not use the official SDK yet.
• Desktop UI uses hand-authored components instead of shadcn scaffolding.
• Source includes deterministic privacy, authority, storage, async, destructive-intent, routing/graph, accessibility, browser, and release controls. Re-run every required gate for the exact candidate; do not reuse a historical pass.
• The live AI-provider smoke runner is opt-in and is not part of the default automated test command.
• Packaged Windows WebView/sidecar, installer/signing, assistive technology, physical devices, and rollback remain qualification obligations until artifact-bound evidence exists. See Frontend qualification.