Files
stack/packages/mosaic
mosaic-coder 7603cb1c10
All checks were successful
ci/woodpecker/pr/ci Pipeline was successful
fix(wake): reconciler fails closed on the observed_seq dual-allocator hazard (#908)
Review of #909 found a blocking correctness defect: the reconciler's
"serialised => safe" claim was FACTUALLY WRONG. observed_seq has two INDEPENDENT
allocators — the detector's private counter ($STATE_DIR/detector/
observed_seq_counter, W4) and the reconciler's store-cursor+offset — so even
STRICTLY SEQUENTIAL calls collide: reconciler enumerates alpha->seq1, beta->seq2
(store observed_seq=2, detector private still 0); the detector's next delta
allocates private 0->1 and enqueues seq1, ALIASING alpha. A consumer acking
CONSUMED 1 then silently DROPS a distinct obligation — the exact G3 swallow the
reconciler exists to prevent, via seq-aliasing. Reproduced with sequential,
never-concurrent calls.

Fix (W5-scoped; full store-side allocator unification stays deferred to #908 —
detector.sh/W4 untouched):
  * Enumeration now FAILS CLOSED. Before writing any observed_seq, the reconciler
    REFUSES (loud, non-zero; the obligation is FLAGGED, never swallowed) when a
    seq it would allocate could be reissued/aliased:
      - if a detector is co-feeding this store (its private state dir exists) —
        ALWAYS refuse (read-only detection; W4 never touched);
      - else require the operator to assert reconciler-SOLE-FEEDER via
        --allow-enumerate / WAKE_RECONCILE_ALLOW_ENUMERATE=1.
    --no-enumerate and `check` never write, so they stay unconditionally safe.
  * Corrected the false "serialised => safe" rationale in the header comment to
    state the real dual-allocator hazard honestly, reference #908, and record the
    operational guard: a detector and the reconciler MUST NOT co-feed one store
    until #908 lands.
  * RED-FIRST test R9 reproduces the serialized-collision scenario (detector
    co-feeding + pre-existing state) and asserts the reconciler FAILS LOUD and
    enqueues NO aliasable seq (depth unchanged). Verified red: removing the guard
    re-enables the collision (R9 depth 0->2) and turns R9 red.

R5/R6 now enumerate under an explicit sole-feeder assertion (no detector
co-feeds those stores). test:framework-shell + shellcheck + firewall all green.

Part of #892

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0158NZqN2n2ymKFeJAZ4GUCb
2026-07-25 20:52:45 -05:00
..

@mosaicstack/mosaic

CLI package for the Mosaic self-hosted AI agent platform.

Usage

mosaic wizard           # First-run setup wizard
mosaic gateway install  # Install the gateway daemon
mosaic config show      # View current configuration
mosaic config hooks list  # Manage Claude hooks

Headless / CI Installation

Set MOSAIC_ASSUME_YES=1 (or ensure stdin is not a TTY) to skip all interactive prompts. The following environment variables control the install:

Gateway configuration (mosaic gateway install)

Variable Default Required
MOSAIC_STORAGE_TIER local No
MOSAIC_GATEWAY_PORT 14242 No
MOSAIC_DATABASE_URL (none) Yes if tier=team
MOSAIC_VALKEY_URL (none) Yes if tier=team
MOSAIC_ANTHROPIC_API_KEY (none) No
MOSAIC_CORS_ORIGIN http://localhost:3000 No

Admin user bootstrap

Variable Default Required
MOSAIC_ADMIN_NAME (none) Yes (headless)
MOSAIC_ADMIN_EMAIL (none) Yes (headless)
MOSAIC_ADMIN_PASSWORD (none) Yes (headless)

MOSAIC_ADMIN_PASSWORD must be at least 8 characters. In headless mode a missing or too-short password causes a non-zero exit.

Example: Docker / CI install

export MOSAIC_ASSUME_YES=1
export MOSAIC_ADMIN_NAME="Admin"
export MOSAIC_ADMIN_EMAIL="admin@example.com"
export MOSAIC_ADMIN_PASSWORD="securepass123"

mosaic gateway install

Runtime launchers

mosaic claude            # Launch Claude Code with Mosaic injection
mosaic yolo claude       # …with --dangerously-skip-permissions
mosaic codex | opencode | pi

mosaic claudex (EXPERIMENTAL)

Runs GPT models inside the Claude Code harness by pointing Claude Code at a local claude-code-proxy that translates the Anthropic Messages API to a ChatGPT-subscription (Codex OAuth) backend. This is not Anthropic Claude — model behavior, tool use, and output quality may differ. Intended for evaluation, not production delivery.

mosaic claudex           # launch (prompts through the proxy readiness gate)
mosaic yolo claudex      # …with --dangerously-skip-permissions
mosaic claudex --print "hello"   # trailing args are forwarded to Claude Code

Prerequisite: the claude-code-proxy binary must be installed and authenticated (claude-code-proxy codex auth …). mosaic claudex runs a preflight that verifies the binary, the OAuth state (triggering a device re-auth if needed), and a trusted local listener before launching; it fails closed if the proxy cannot be brought up with a verified identity.

Isolation (never touches your real Claude state). claudex always launches against an isolated CLAUDE_CONFIG_DIR (default ~/.config/mosaic/claudex/home). The ambient CLAUDE_CONFIG_DIR is deliberately ignored, and a guard proves the resolved dir can never be — or live under — the real ~/.claude. A claudex session therefore cannot mutate your normal Claude Code config.

No token leakage. claudex never reads the proxy's credential file. Claude Code is handed only ANTHROPIC_AUTH_TOKEN=unused pointed at the loopback proxy; the entire credential-bearing env family (ANTHROPIC_*, AWS_*, GOOGLE_CLOUD_*, GOOGLE_APPLICATION_CREDENTIALS, *_TOKEN, *_KEY, *_SECRET, …) is stripped from the composed environment. The Bedrock/Vertex routing switches (CLAUDE_CODE_USE_BEDROCK, CLAUDE_CODE_USE_VERTEX, and the _SKIP_*_AUTH pair) are force-removed regardless of value — otherwise their mere presence would route Claude Code to the real Anthropic API via AWS/GCP and bypass the proxy. The proxy holds the real OAuth credential.

Model tiers (override via env).

Tier Env var Default
primary (opus/sonnet) ANTHROPIC_MODEL gpt-5.6-sol
small/fast (haiku) ANTHROPIC_SMALL_FAST_MODEL gpt-5.6-luna

Operator-provided values win over the defaults. Additional overrides: MOSAIC_CLAUDEX_CONFIG_DIR (isolated config dir), ANTHROPIC_BASE_URL (proxy endpoint).

Hooks management

After running mosaic wizard, Claude hooks are installed in ~/.claude/hooks-config.json.

mosaic config hooks list              # Show all hooks and enabled/disabled status
mosaic config hooks disable PostToolUse  # Disable a hook (reversible)
mosaic config hooks enable PostToolUse   # Re-enable a disabled hook

Set CLAUDE_HOME to override the default ~/.claude directory.