Provider-neutral coding agent and multi-agent orchestration toolkit.

OMK routes work across models, constrains tools and paths, verifies outcomes, and keeps replayable evidence.

OMK is a verified, provider-neutral control plane for Codex, Claude Code, OpenCode, and local coding agents.

It scopes parallel work into bounded DAG lanes, blocks completion without fresh evidence, and records replayable artifacts for review and recovery.

Ten short captures of the control plane's main workflows.

• Execution scope — resource claims, owned paths, bounded parallel lanes
• Completion — declared predicates and fresh verification
• Evidence — commands, exit status, workspace state, receipts
• Recovery — replayable session state and repair tooling
• Providers — one operator model across supported coding agents

npm install -g open-multi-agent-kit --ignore-scripts
omk --version
omk

Or without a global install:

npx --ignore-scripts open-multi-agent-kit

The open-multi-agent-kit package ships OMK.

The OMK//CONTROL startup surface is the default operator view. The header reads omk v<package.version> · OMK//CONTROL, using the installed workspace package version as its source of truth.

!omk plan turns a fuzzy objective into a bounded DAG: owned paths, ordered waves, and an acceptance predicate attached to every node before a single line of code is written.

The Correctness Wall intercepts writes and runs acceptance predicates. A red predicate blocks completion — a green-looking reply is never a release signal.

Every verified run produces a receipt: commands, exit codes, workspace state, evidence digest, and timestamps. Receipts are inspectable artifacts, not marketing claims.

omk session doctor detects unterminated turns and orphan results, then plans a dry-run repair against the tamper-evident replay ledger. An interrupted run is a recoverable state, not a loss.

OMK is provider-neutral. The underlying agent can be Codex, Claude Code, OpenCode, or a local model; the execution and evidence model stays consistent.

Providers stay interchangeable. The routing layer picks the best arm for the task, but the control plane never changes when you swap models.

OMK is not a security sandbox for arbitrary hostile code by default. Supported verification and sandbox modes are documented explicitly. Any run without the required evidence is labeled UNVERIFIED.

See containerization.md for sandbox patterns: OpenShell, Gondolin micro-VM, and plain Docker.

OMK packages distribute skills, extensions, prompts, and themes through one control plane. Build a package once, install it through omk, pin it, scope it to a project when needed.

# Global, pinned OMK package
omk install npm:some-omk-package@1.2.3

# Project-local, pinned Git package
omk install -l git:github.com/example/omk-package@v1.2.3

# Inspect and control installed resources
omk list
omk config
omk update --extensions

A skills-only package is an ordinary OMK package:

{
  "name": "omk-workflows",
  "keywords": ["omk-package"],
  "omk": {
    "skills": ["./skills"]
  }
}

Use the minimum necessary skills per turn — usually one to three. A skill is loaded when it earns its place in the task, not because it happens to be installed.

npm install omk-agent-core   # Agent runtime
npm install omk-ai           # Multi-provider LLM API
npm install omk-tui          # Terminal UI

AdaptOrch MCP is a separate, proprietary reliability-kernel service (not part of this monorepo) that OMK can route orchestration tasks through: topology-aware DAG routing, multi-model synthesis, and consistency verification. Backed by a published paper (arXiv:2602.16873).

The adaptorch and adaptorch-prod MCP servers plus the adaptorch-route and adaptorch-synthesize skills ship in OMK's default execution preset. Actually invoking AdaptOrch still requires an ADAPTORCH_CONTROL_PLANE_TOKEN.

This is distinct from packages/adaptorch-wpl in this monorepo, the stable Work Packet Loop package shipped as a runtime dependency of open-multi-agent-kit since v0.91.0.

npm ci --ignore-scripts  # Install the locked dependency graph
npm run build            # Build all packages
npm run check            # Lint, format, and type check
npm test                 # Run the hermetic default test suite
./omk-test.sh            # Run OMK from sources

• Direct external dependencies are pinned to exact versions.
• .npmrc sets save-exact=true and min-release-age=2.
• package-lock.json is the dependency ground truth.
• npm run check verifies pinned direct deps and the generated shrinkwrap.
• The published CLI includes npm-shrinkwrap.json to pin transitive deps.
• CI installs with npm ci --ignore-scripts; scheduled audits run npm audit.

See CONTRIBUTING.md for contribution guidelines and development.md for project setup.

• Read the documentation
• Browse all public Skills
• Release notes for v0.95.0

• Added explicit multi-account subscription authentication: /login now opens an account picker for configured OAuth providers, offers a separate Add another account action, captures ChatGPT/Claude/Google email identities, and pins requests to the selected account without silent rotation or failover.
• Added multi-provider quota meters to the pinned STATUS RAIL: every configured Codex, Claude, Kimi Code, and GLM/ZAI subscription is shown together with independent windows and reset countdowns. Codex polling passively merges official x-codex-primary-*, x-codex-secondary-*, and codex.rate_limits signals when they are present. Claude Code's passive anthropic-ratelimit-unified-* headers preserve recent 5-hour/7-day values; when the usage endpoint is rate limited and the snapshot is incomplete, an account-scoped, hourly-capped one-token Haiku quota check mirrors Claude Code's startup fallback. Model Studio Token Plan is identified as QWEN TOKEN PLAN with console-only quota because its official usage endpoint requires an Alibaba Cloud console session rather than the plan API key; Qwen OAuth/Grok remain explicit quota API unavailable entries.

• Removed the persistent GitHub star prompt from interactive startup. The /star command remains available as an explicit, on-demand repository shortcut and no longer writes a tracking flag to settings.
• Anthropic-bound images now default to a 1,900 px longest edge and pass through a final header-level size guard, preventing sticky oversized-image request failures from clipboard and extension paths.
• Cancelling manual compaction now clears the transient compaction notice and restores queued input without duplicating messages.
• Refreshed built-in model metadata via omk-ai so OpenCode Zen (opencode) gains the kimi-k3 entry that OpenCode Go (opencode-go) already had, with current provider handoff coverage for both catalogs.
• glm-5.2 on OpenCode Zen/Go now exposes the xhigh/max thinking levels in the thinking-level selector (Ctrl+T / /thinking); it was previously capped at high because top-tier levels require an explicit thinkingLevelMap entry to appear.
• Auto-compaction now fails fast on insufficient-balance 429 responses instead of repeating the same quota error through every retry.

Release notes live in RELEASE_NOTES_v0.95.0.md.

• Persistent GitHub star nudge on interactive startup: first installs and anyone who has not confirmed a star keep seeing a nag banner every launch until they star https://github.com/dmae97/omk and run /star (writes global githubStarred: true in ~/.omk/agent/settings.json). /star reset brings the nag back. Project settings cannot silence it.

Release notes live in RELEASE_NOTES_v0.94.1.md.

• diagnostics is now a default-active tool for new sessions (LLM-callable alongside read/bash/edit/write). It stays registered-but-inactive for sessions that pin their own tool list; opt out per session via activeToolNames or excludedToolNames.
• Persistent skill-catalog cache (src/core/skills-catalog-cache.ts): per-dir fingerprint walk (readdir/stat only) gates a JSON catalog at <agentDir>/cache/skill-catalog-v1.json. Repeat session starts skip all SKILL.md reads on unchanged trees (measured on this host: 105 ms cold → 41 ms warm for the full scan). Any add/edit/delete under a scanned tree invalidates exactly that dir; corrupt cache degrades to a clean miss. Atomic tmp+rename writes.
• Hermetic test environment (test/setup-env.ts): machine-level OMK_* and provider credential variables are scrubbed before every worker, so test results no longer depend on the developer shell (safety-gate suites saw env-disabled gates; live e2e suites ran against expired credentials instead of skipping). LIVE_E2E=1 keeps provider keys when running the live suites on purpose.
• diagnostics tool (src/core/tools/diagnostics.ts): compiler-backed diagnostics via the project's own checkers — tsc --noEmit, pyright/ruff, go vet, cargo check — normalized to SEVERITY path:line:col message, per-language fail-soft (skipped instead of tool errors), 5 s TTL cache, 50-item cap, path/language auto-detect. Registered in createAllToolDefinitions and exported from the SDK (createDiagnosticsTool, createDiagnosticsToolDefinition).
• Interactive sandbox promotion: session.setBashSandboxMode("audit" | "enforce" | "off") switches the session sandbox at runtime (next spawn), with a sandbox_audit mode-change ledger entry; session.bashSandboxMode reads the effective mode. SessionBashRuntime.setSandboxMode backs it.
• Default-on bash sandbox (opt-out). Session bash now carries a default audit-mode sandbox preflight (workspace-write rooted at the session cwd, OS temp dir as extra write target). Spawns stay unwrapped but every decision lands in the replay ledger as a sandbox_audit event — a tamper-evident trail no other harness ships. OMK_BASH_SANDBOX=enforce activates the real OS backend (macOS sandbox-exec / Linux bwrap, auto-detected) and fails closed when unavailable; =0 disables. New onSpawnDecision observer on BashSandboxPreflight, plus createWorkspaceSandboxPolicy() / resolveBashSandboxMode() SDK exports.
• Git-aware verified-bash scope. Session bash receipts now bind the git toplevel plus the sorted dirty set (staged/modified/untracked, capped at 32 paths, 1 s TTL cache) instead of an empty artifact set, so captureWorkspaceFingerprint records HEAD and a scope-limited dirty digest. Exported as resolveSessionWorkspaceScope().

• API provider registry is now a process-wide singleton (globalThis-anchored in omk-ai/api-registry). Symlinked workspace dist copies consumed natively and the same files inlined by vite-node used to keep separate registries, so registerFauxProvider (and any runtime registration) was invisible to streamers resolving through the other copy — surfacing as "No API provider registered for api: ..." in agent loops. Also removed a stale nested packages/agent/node_modules/omk-ai copy (0.92.0) that shadowed the workspace build.

• Extracted SessionCompactionService (src/core/session-compaction-service.ts): the compaction state machine — capture/lock, barrier evaluation, emergency tail repair, provenance capture, transaction begin, envelope commit — moved out of AgentSession (5,271 → 4,911 lines), which now delegates through thin one-line wrappers. Transaction symbols import from compaction/transaction.ts directly so the compaction/index.js vi.mock pattern in suites keeps working.

• Extracted SessionBashService (src/core/session-bash-service.ts): the full bash surface — executeBash (prefix/loadout/safety-floor/headless gate), recordBashResult with the streaming-deferral queue, abortBash, flushPending — moved out of AgentSession, which now delegates one line each. Ordering contract (queue while streaming, flush on turn end) is pinned by the bash-persistence suite.

• Extracted SessionBashRuntime (src/core/session-bash-runtime.ts) from AgentSession: verified-evidence executor, default sandbox preflight (audit/enforce), git-aware workspace scope, and the receipt-bound bash orchestration (executeVerified) now live in one lazily-initialized unit; the session delegates. No behavior change.

• Verified bash is default-on (opt-out). When an AgentSession has a replay ledger, LLM-callable bash and interactive/RPC executeBash bind through executeVerifiedBash (executor: "bash-tool", receipts under <sessionFile>.evidence). Set OMK_VERIFIED_BASH=0 for the legacy unverified path. Adapter gains env/onData fan-out and createVerifiedBashOperations() so session PI_* env and live streaming stay intact without import cycles. See SDK — Evidence and Environment Variables.

Release notes live in RELEASE_NOTES_v0.94.0.md.

MIT