open-multi-agent/CHANGELOG.md

Changelog

Unreleased

1.21.0 - 2026-09-25

Added

• Added runImage() for image generation and editing across an ordered chain of image models, with per-model retries that honor Retry-After, fallback on non-retryable failures, an optional validate check on each returned image, and a record of every provider call through result.attempts and onAttempt.
• Added the ImageModelAdapter interface and four built-in adapters that call provider HTTP APIs directly and honor egressPolicy: OpenAIImageAdapter (OpenAI Images API and OpenAI-compatible endpoints), SeedreamImageAdapter (Seedream on Volcengine Ark), OpenRouterImageAdapter (OpenRouter /images, sending input images as input_references), and BlackForestLabsImageAdapter (FLUX models, submit/poll/download).
• Added ImageModelError with normalized ImageModelErrorType values; isRetryableError() returns its retryable flag, its optional params carries details of already-accepted provider work such as a task ID and cost into the failed attempt record, and ImageCallOptions.maxRetryAfterMs tells an adapter that waits internally the longest Retry-After runImage() allows.
• Image adapters classify safety or moderation rejections, including upstream rejections forwarded in error.metadata, as non-retryable content_policy, fail rather than drop input they cannot use, and report the media type read from the returned image bytes.

Changed

• Enabling thinking without budgetTokens now sends adaptive thinking on Claude Opus 4.7, Sonnet 5, and later, which previously rejected the request with HTTP 400; pre-adaptive models keep the 1024-token budget default.
• The task profiler omits temperature for Claude models that reject non-default sampling parameters, avoiding HTTP 400 routing failures on Opus 4.7, Sonnet 5, and later.
• The bash tool description now states that it is not confined to the agent's working directory, prefers the file tools, grep, and glob for file work, and reports a non-zero exit code; built-in tool descriptions now match glob's path output, and the synthesis prompt no longer carries an unscoped JSON-only instruction.
• create-oma-app is republished as a patch so its templates pin the new core version exactly; otel is unchanged and keeps its semver range against core.

Fixed

• Enabling thinking without budgetTokens no longer produces HTTP 400 on current Claude models (Opus 4.7, Sonnet 5, and later); those models now receive adaptive thinking instead of an explicit budget.
• The task profiler no longer sends temperature to Claude models that reject non-default sampling parameters, fixing HTTP 400 routing failures on Opus 4.7, Sonnet 5, and later.

Compatibility

• No breaking changes: core public API changes are additive exports only, no export was removed, renamed, or narrowed, and no engines floor or published dependency major changed; isRetryableError() gains an answer for ImageModelError while all pre-existing error classes keep their prior answers, and no import path changes.
• Callers that pass an explicit thinking.budgetTokens on Claude Opus 4.7 or Sonnet 5 must still omit it; passing it explicitly still emits budget_tokens and is rejected with HTTP 400.
• Callers that relied on the implicit 1024-token thinking budget on Claude 4.6-generation models now get adaptive thinking and should re-check maxTokens and thinking-cost assumptions.
• If a model ID is routed through a proxy, alias, or non-canonical suffix, verify thinking and temperature behavior against that endpoint before upgrading: the legacy-model list in anthropic-models.ts is closed, and unknown IDs default to adaptive thinking and no temperature.
• For the new image adapters, supply credentials via apiKey or the environment (OPENAI_API_KEY, OPENROUTER_API_KEY, ARK_API_KEY, BFL_API_KEY); providerOptions keys that carry the request itself (prompt, model, n, input image fields, mask, response_format where applicable) are rejected with a TypeError at construction. A request's size overrides a providerOptions size default; Seedream watermark is set on the adapter, with providerOptions watermark overriding the adapter's watermark option. It has no per-request override.
• An egressPolicy allowlist used with the Black Forest Labs adapter must include the poll and download hosts in addition to the API origin.
• create-oma-app republishes as a patch so its templates pin the new core version exactly; already-generated projects keep working and can upgrade core independently.
• otel stays at its current version with its existing semver range against core; it is not republished and remains compatible with the new core.

1.20.0 - 2026-09-18

Added

• Added an optional positive-integer GateThreshold minSamples guard. A threshold below the minimum reports insufficient_samples with the observed count as actual and the configured minimum as limit; score metrics compare against the selected aggregate scoredCount, passRate compares against passSampleCount, and tag-scoped thresholds use the tag aggregate own counts.
• Added ScorerAggregate passSampleCount, the number of scored records that define pass. It is emitted whenever passRate is emitted, including on tag aggregates.

Changed

• Baseline regression comparisons are skipped with a warning when either the current or baseline aggregate holds fewer than minSamples samples; this can turn a previously reported regression failure into a warning.
• A report or baseline written before passSampleCount existed fails closed at zero samples for passRate guards rather than skipping the guard.
• Updated the shipped README and npm package description to lead with OMA as the project name and reposition around owning, approving, and auditing.

Compatibility

• Omitting minSamples preserves previous threshold and regression behavior, and the report schema version remains unchanged, so existing gate policies and reports need no migration to keep prior behavior.
• Before enabling minSamples on a passRate threshold, regenerate baselines with a core version that emits passSampleCount; otherwise that regression comparison skips with a warning, and a passRate guard evaluated against an old current report fails closed with actual zero.
• Adding minSamples changes gate outcomes in both directions: a short baseline can downgrade a regression failure to a warning, while a short current report still fails its own threshold with insufficient_samples.
• A previously ignored minSamples key with a non-positive or non-integer value now throws minSamples must be a positive integer; remove or correct such keys in existing gate policies.
• If code exhaustively switches on GateFailure kind, add an insufficient_samples case. The new kind carries the observed sample count as actual and the configured minimum as limit, with optional scorer, metric, and tag coordinates.
• Report JSON now includes passSampleCount on every aggregate that carries passRate, including tag aggregates. Update snapshots, hashes, or strict validators; the unchanged schema version cannot distinguish the two shapes.
• Do not rely on an older core to enforce a policy that declares minSamples, because that core strips the unknown key and silently skips the guard; pin the gate execution side to a core version that supports it.
• No Node or dependency action is required: the core manifest change is a description-only edit, and the otel and create-oma-app workspaces have no changes in this range.

1.19.0 - 2026-09-11

Added

• Added an opt-in authoritative run store through OrchestratorConfig.runStore and RunTasksOptions.runStore; it gives a logical run a durable lifecycle record, an execution lease, and a monotonically increasing fencing token so one worker at a time advances the run and a worker that was taken over cannot write after the takeover.
• Added the RunStore interface, the MemoryStoreRunStore adapter over any MemoryStore with compare-and-set, and the RunLedger and RunLeaseHandle surface for reading, cancelling, and resuming a run from outside its worker.

Changed

• TeamConfig.maxConcurrency now bounds the agent pool for that team's runs instead of being accepted and ignored. It intersects with OrchestratorConfig.maxConcurrency, and the smaller value wins, so a team can narrow its own pool but never widen it past the orchestrator ceiling. A team that set a cap below the orchestrator's was previously running at the orchestrator value and now runs at its own lower value; raise or remove the team cap to keep the previous throughput. A cap that is not an integer of at least 1 is reported as an INVALID_TEAM_MAX_CONCURRENCY warning on onProgress and the orchestrator value applies.
• With a run store configured, execution-ownership and run-lifecycle writes fail closed instead of following the best-effort semantics of checkpoint and telemetry writes: a fenced-out checkpoint write never reaches the store, a lost lease stops the run at the dispatch gate, and a worker that lost its lease reports the fence failure rather than success. Runs with no run store configured are unchanged.
• The oma CLI default model for provider: 'deepseek' is now deepseek-flash (DeepSeek-V4.1-Flash), replacing deepseek-v4-flash. The retired name still resolves, and library users that pass an explicit model are unaffected. Provider docs, adapter JSDoc, the scaffolder env template, and the DeepSeek examples moved to the same name.
• The DeepSeek adapter JSDoc no longer states that thinking runs at high effort. The adapter forwards whatever effort the caller passes and sets none itself; runtime behavior is unchanged.
• Corrected published JSDoc that misstated runtime behavior, including onAgentStream coverage in runTasks(), done stream-event payloads per layer, retryable-error classification, and the RedactingStore example constructor call.
• The delegate_to_agent error returned when the pool has no free slot now points at both ceilings: "Increase maxConcurrency on the orchestrator or the team" rather than the orchestrator alone. Callers that match on that string should update it.
• The @open-multi-agent/core package description on npm was rewritten to lead with running in your own environment, approval gates, and a verifiable record of every run.
• LICENSE now names Shenzhen YuanASI Technology Co., Ltd. alongside the open-multi-agent contributors, package READMEs close with a maintainer line, and the author field in the published package.json files points at YuanASI. The MIT terms are unchanged.

Compatibility

• Runs that do not configure a run store keep the existing single-process checkpoint and restore behavior. restore() is the resume command either way; with a run store it additionally makes a suspended record eligible and takes the lease before reconciling the approval ledger.
• A team that set maxConcurrency below the orchestrator value previously ran at the orchestrator value and now runs at its own lower cap. Raise or remove the team cap, or raise OrchestratorConfig.maxConcurrency, to keep the previous parallelism; leave the field unset to keep the old behavior exactly.
• Non-integer or less-than-1 team caps no longer pass through silently. They emit an INVALID_TEAM_MAX_CONCURRENCY warning on onProgress and fall back to the orchestrator value. This is a new code on the existing warning event; no event type was added.
• The old deepseek-v4-flash name still resolves, and library callers that pass an explicit model are unaffected. Cost maps, telemetry filters, or snapshot assertions that key on the model string should be reviewed because the CLI default now sends deepseek-flash.
• No existing export was removed or renamed and every new config field is optional, so existing code keeps compiling and running unchanged unless it opts into the run store or relies on ignored team concurrency caps.
• The otel package changed only in shipped-file metadata; it has no public API or runtime change.
• The .env.example that create-oma-app scaffolds now names deepseek-flash in the commented-out OMA_MODEL line of its DeepSeek section; existing generated projects are not rewritten by this release.

1.18.0 - 2026-09-04

Added

• Added VideoBlock with base64 or absolute HTTP(S) URL sources to the public content-block API, serialized as an OpenAI-compatible video_url part for MiniMax-compatible chat.
• Added UnsupportedContentBlockError for adapters that cannot represent a content block; unsupported video input is terminal and excluded from retry classification.
• Tool input and output are now recorded on execute_tool v2 spans as oma.tool.input and oma.tool.output when the trace capture policy opts in.
• Added contentCapture: { mode: 'upstream-policy' } to @open-multi-agent/otel 0.1.3, forwarding core's opt-in tool input/output attributes.
• Evaluation judgePrompt can now return per-judge structured LLMMessage[] input in addition to the existing plain-text form.

Changed

• Context-summary preparation now strips video blocks like image blocks, and token estimation sizes inline video as rich media so compaction triggers correctly.
• The @open-multi-agent/otel contentCapture option is widened from only 'disabled' to 'disabled' or 'upstream-policy'; default behavior is unchanged.
• Evaluation judgePrompt is now invoked once per judge instead of once per score() call, so a quorum calls it once for each member. Callers whose judgePrompt is costly or has side effects will see the additional calls.

Fixed

• Fixed v2 trace sinks silently dropping tool input/output that the legacy onTrace callback delivered; recording is now gated by an explicit capture policy with default behavior unchanged.

Compatibility

• Core remains backward-compatible: all reviewable executable changes are additive exports, an additive ContentBlock union member, trailing optional parameters, or opt-in trace attributes; no published dependency/peer ranges or engine floors changed.
• TypeScript consumers doing exhaustive matching over ContentBlock should add a 'video' case or default when upgrading because the union is widened.
• No migration is required for default callers: v2 tool I/O capture remains off, otel contentCapture remains disabled by default, and existing retry classifications are unchanged.
• To retain tool arguments/results previously delivered by the legacy onTrace callback, enable the core capture policy and the otel upstream-policy mode; both opt-ins are required, and the mode needs @open-multi-agent/otel 0.1.3 or newer.
• Eval callers using judgePrompt's structured return path must append buildStructuredOutputInstruction themselves; the existing (context) => string signature is unchanged and its result is still wrapped by the text template.
• Video sources must be absolute HTTP(S) URLs or raw base64 with a parameterless MIME type; adapters without a video mapping fail fast with the terminal unsupported-content-block error rather than retrying.

1.17.0 - 2026-08-28

Added

• Added an opt-in append-only run event journal (RunEvent vocabulary, InMemoryRunJournal, JsonlRunJournal) that records every adapter call and the per-block lineage of what the model saw.
• Added verifyRun() to cold-verify a finished run journal, proving model-visible blocks are reproducible from the events they name.
• Added journal-watermarked checkpoint schema v5 for journaled runs; v1–v4 snapshots keep loading and unjournaled runs keep writing byte-identical v4 snapshots.
• Added JournalLineageError and the enforceLineage switch; built-in context strategies now emit context/replace events for every persistent rewrite.
• Added per-call journal options to runAgent/runTasks options and OrchestratorConfig, off by default with best-effort appends that never fail the run.

Changed

• Journaled restores now replay the journal tail after the stalest per-task watermark, re-anchoring in-flight state and replaying committed tool results that reached the journal but not the checkpoint.

Fixed

• Aligned JSON Schema required fields with Zod validation.

Compatibility

• Run journaling and checkpoint v5 are opt-in and additive: no public export was removed or narrowed, and a run that leaves journaling off still writes byte-identical v4 snapshots.
• Generated JSON Schema for tool parameters now derives required from Zod's own optionality check rather than the declared wrapper type. Three property shapes change as a result: .nullable() becomes required, while z.any() and z.unknown() stop being required. Built-in tools are unaffected, and MCP tools forward the server's own schema, so this reaches custom tools that use those shapes in a parameter position.

1.16.1 - 2026-08-21

Added

• Export eight previously internal public types from the package root (ThinkingConfig, ToolResultMetadata, ProcessAgentBackendConfig, ProcessBackendInputMode, CostEstimateContext, MessageBusSnapshot, MessageSnapshot, MessageReadStateSnapshot)
• Document DeepSeek V4-Pro-0813 / V4-Flash-0731 GA snapshots and default thinking at high effort

Changed

• Accepted verify revisions now refresh structured output together with the revised answer instead of leaving stale parsed output
• Slim core package READMEs, move contributor credits to CONTRIBUTORS.md, and add LLM egress policy / shell executor coverage to the Chinese READMEs

Compatibility

• All public API changes are additive: eight new type-only exports from the package root
• No public export was removed or narrowed, so existing consumers can upgrade without code changes

1.16.0 - 2026-08-16

Added

• Add opt-in EgressPolicy (offline or allowlist) that restricts framework-owned LLM network requests and intersects run-, agent-, and orchestrator-level scopes.
• Add pluggable ShellExecutor / LocalShellExecutor so callers control where the built-in bash tool runs, exposed via the new @open-multi-agent/core/shell subpath export.
• Add StructuredOutputValidationError, which marks persistent schema failures after the built-in corrective attempt as terminal for task-level retry.
• Add EgressPolicyError with stable codes (INVALID_EGRESS_POLICY, EGRESS_POLICY_DENIED, EGRESS_POLICY_TARGET_UNRESOLVED, EGRESS_POLICY_UNSUPPORTED) for disallowed or unenforceable LLM egress.

Changed

• The built-in bash tool now executes through a configurable shell executor with a fail-closed backstop and 127 exit code on execution failure.
• LLM adapters now resolve provider default base URLs and enforce egress policy before opening framework-owned requests; policy use is opt-in and preserves legacy unrestricted behavior.
• Task retry now respects a result's stable errorInfo.retryable: false classification.

Fixed

• Structured-output agents no longer report a late valid response as success after their whole-run timeout fired.
• Task retry no longer reruns validation, cancellation, and budget failures when the raw in-process error is unavailable.

Security

• Add framework-owned LLM egress restriction (EgressPolicy) that can confine provider transports to loopback or an explicit origin allowlist.

Compatibility

• All core additions are backward compatible: new exports, optional config fields, and new error classes; no public export was removed or narrowed and the Node engines floor is unchanged.
• Egress policy is opt-in: omitting egressPolicy on run/agent/orchestrator config preserves legacy unrestricted LLM egress behavior.
• Shell executor is opt-in: AgentConfig.shellExecutor / OrchestratorConfig.defaultShellExecutor default to LocalShellExecutor for the built-in bash tool; callers that invoked bashTool.execute() directly now own the executor lifecycle and may observe results with exitCode 127 instead of a thrown tool error when execution fails.
• StructuredOutputValidationError and EgressPolicyError are now classified as non-retryable by isRetryableError and mapped to terminal failure states; task-level retry behavior for such failures changes.
• New public surface added in this release: ./shell subpath export (LocalShellExecutor, ShellExecOptions, ShellExecResult, ShellExecutor), StructuredOutputValidationError, EgressPolicyError, EgressPolicy, and egressPolicy/shellExecutor config fields.

1.15.0 - 2026-08-09

Added

• Durable approval gates can suspend plan, round, task-dispatch, and tool-call decisions, persist content-bound pending requests and first-wins reviewer decisions in a checkpoint MemoryStore, and resume only the approved plan, task, or validated tool invocation. Missing state, stale hashes, schema drift, and tampering fail closed, while derived execution receipts carry the review evidence without making telemetry the authoritative record. New checkpoint writes use schema version 4 to preserve approval continuation state.
• Checkpoint recovery preserves in-flight built-in runner messages, turn and token accounting, pending tool calls, and independently committed tool results. Restore replays committed results without re-executing their tools, reruns only missing calls, and exposes the stable model-issued toolCallId to tools and per-call gates for external idempotency.
• Agent.run(), Agent.stream(), and OpenMultiAgent.runAgent() accept complete structured message histories. Persistent Agent.prompt() conversations can accept one structured user turn, including image input, with runtime validation, defensive copies, and full-message access in beforeRun.
• Rich tool results separate application-owned ToolResult.data from validated, model-visible text, image, and file modelOutput. Built-in adapters and MCP map supported content explicitly, while runner context, checkpoints, recovery, progress, evaluation, and privacy-safe trace summaries preserve the richer result.
• AgentConfig.history restores persisted conversation messages into a new Agent so subsequent prompt() calls can continue an earlier conversation.

Compatibility

• Existing string inputs, string/error tool results, and compression markers keep their previous behavior. Structured input and rich tool output are additive; adapter- and model-specific media limits still apply.
• Checkpoint readers remain compatible with version 1, version 2, and version 3 snapshots. OMA cannot atomically commit an arbitrary external side effect together with its checkpoint write, so consequential integrations still need external idempotency for that crash window.
• runTeam() and runTasks() remain text-only. Process and ACP backends remain task-grained and reject structured input rather than silently dropping it.
• Existing approval callbacks can continue returning allow or deny. Suspension is additive; task gates with live verification wiring, standalone runAgent(), the simple-goal shortcut, process backends, and ACP backends do not expose every resumable private tool-loop path.
• @open-multi-agent/otel@0.1.1 is not republished. Its @open-multi-agent/core@^1.11.0 dependency remains compatible with core 1.15.0.

1.14.0 - 2026-08-01

Breaking changes

This release shipped as a minor version, so a caller on a ^1.x range receives it without an explicit upgrade step. npm reports an engines mismatch as an EBADENGINE warning rather than an install failure, so a Node 18 project installs 1.14.0 successfully and then fails at run time.

• Node.js 20 or newer is now required. The engines floor moved from 18 to 20 across @open-multi-agent/core, @open-multi-agent/otel, and create-oma-app. Node 18 reached end of life on 2025-04-30.
• The bundled openai dependency moved from v4 to v6. A project that also depends on openai directly resolves a nested second copy until it moves to v6 as well.
• Plans that were previously accepted can now fail at validation time. Invalid task dependency graphs, unsatisfiable task requirements, and invalid coordinator plans are rejected before execution instead of running partially. Each of these surfaces a defect that was previously silent, so correct graphs and rosters are unaffected. The individual entries are under Changed and Fixed below.

Added

• Adaptive plan recovery lets a run revise the not-yet-executed part of its task graph. Opt in with recovery.mode: 'repairable', then supply a Replanner or an onTaskOutcome callback that proposes an append-only PlanPatch. Patches are validated for agent eligibility, limits, task states, references, and the resulting DAG, gated through the optional onPlanPatch approval, and applied atomically at a task-outcome barrier before downstream dispatch or failure cascade. Revision history is exposed in results, progress events, and observability spans.
• Hybrid semantic execution routing supplements the deterministic router with a single structured semantic assessment. Opt in with executionRouting: { strategy: 'hybrid' }. The release adds a provider-neutral TaskProfiler interface, a built-in LLMTaskProfiler, a strict task-profile schema, typed routing failures with timeout and fallback metadata, and semantic-routing observability.
• DeepSeek V4 Flash reasoning controls. AgentConfig.thinking.enabled now maps to DeepSeek's native thinking.type, and thinking.effort accepts the DeepSeek-only value 'max' without forwarding it to OpenAI, Azure OpenAI, or GitHub Copilot.
• validateTaskRequirements is exported for callers that want to check task requirements against a roster before dispatch.
• New typed errors are exported for the failure modes above: InvalidTaskRequirementsError, UnsupportedToolCallError, RoutingDeclarationRequiredError, RoutingProfilerFailedError, and RoutingTimeoutError.

Changed

• On the OpenAI v6 SDK, user aborts are now classified as cancellation rather than as a retryable failure, and unsupported custom tool calls are rejected explicitly instead of being passed through.
• Task requirements are enforced as global hard constraints. A task whose requirements no agent satisfies is now rejected rather than assigned to an ineligible agent.

Fixed

• Invalid task dependency graphs are rejected up front instead of executing a partially valid plan.
• The coordinator fails closed on an invalid plan rather than continuing with a plan it could not validate.

Compatibility

• Automatic runTeam() routing remains deterministic. Hybrid semantic routing is opt-in through executionRouting.strategy and does not change existing runs.
• Adaptive plan recovery is opt-in. Task graphs stay fixed unless recovery.mode selects 'repairable'.
• Adaptive recovery adds a version 2 task-queue snapshot carrying plan-revision history. TaskQueue.fromSnapshot() still accepts version 1 snapshots, so checkpoints written by earlier releases remain restorable.
• Every public export from 1.13.0 is still exported. New result and configuration fields remain optional, so existing callers and serialized results continue to type-check.

1.13.0 - 2026-07-24

Added

• Execution routing can now be selected explicitly with mode, customized through ExecutionRouter, or left to the built-in DeterministicRouter. Every runTeam() topology choice exposes a structured routing decision and trace linkage.
• Structured governance declarations support required or preferred roles, ordered review paths, budget-aware degradation, post-execution conclusions, and privacy-preserving execution receipts.
• Consequential tools can be declared through ToolDefinition.consequential. Undeclared runs expose a machine-readable disclosure flag and can opt into confirmation through the existing onToolCall gate.
• Model routes can declare ordered fallback routes for retryable worker provider failures.
• Agents and tasks can declare structured capabilities and hard requirements. The scheduler adds capability-match and weighted composite strategies, structured warnings, and optional strict assignee validation.
• TeamRunResult.taskResults preserves unmerged results by task ID. Explicit tasks can choose raw, structured, or combined dependency payloads and attach bounded role/provenance metadata.
• OrchestratorConfig.onTaskDispatch(task) provides a native per-task pipeline approval gate. It is mutually exclusive with onApproval.
• The offline Run Viewer surfaces execution-routing decisions, and Evaluation includes a language-neutral routing-stability gate.

Changed

• Task DAG execution is event-driven by default. A downstream task now starts when its dependencies are satisfied instead of waiting for unrelated tasks from the same ready set.
• Progress events from independent DAG branches may interleave instead of arriving in round-sized groups. Consumers should correlate events by task ID and use task status plus dependsOn rather than adjacency to derive state.
• Unassigned tasks are scheduled one ready task at a time against the current DAG snapshot. Dependency-aware ready-set ordering and existing strategy eligibility/fallback contracts are preserved.
• Abort, budget exhaustion, and task-dispatch approval rejection now stop new dispatches, drain in-flight work, and then skip remaining tasks.
• Automatic execution routing recognizes structured Chinese, Japanese, and Korean goals and uses script-aware information length instead of relying on English-only word patterns and raw character count.

Fixed

• CJK keyword extraction and zero-score fallback no longer select an ineligible agent or lose a valid keyword-based match.
• Governed planOnly runs validate and return the declared role DAG without executing agents.
• Explicit execution modes, governance floors, and per-run token/cost ceilings now resolve through a documented precedence order and disclose overrides or budget degradation instead of silently changing topology.

Compatibility

• Configuring the existing onApproval callback automatically retains legacy round scheduling and callback semantics. A separate legacyBatchScheduling flag is not provided because onApproval already selects that compatibility path.
• Custom UIs that depend on round-grouped progress timing can temporarily configure onApproval: async () => true; event-driven consumers should migrate to task-ID correlation.
• Raw dependency output remains the default. Structured dependency handoff, governance declarations, consequential confirmation, and custom execution routing are opt-in.
• New result fields remain optional in public TypeScript interfaces so older serialized results and caller-authored fixtures continue to type-check.
• @open-multi-agent/otel@0.1.0 is not republished; its @open-multi-agent/core@^1.11.0 dependency remains compatible with core 1.13.0.

1.12.1 - 2026-07-20

Evaluation V1 (#403–#409)

• New @open-multi-agent/core/eval and /eval/file entry points.
• Define reusable Scorer implementations and versioned EvalSet fixtures.
• Run deterministic offline evaluations with reports, stores, gates, CLI commands, and reference scorers.
• Sample and score production runs online without changing the business result.

Offline Run Viewer (#394, #411)

• Open saved results and traces locally without a running service.
• Inspect task-level model, provider, token, and cost information through the viewer and CLI.

Per-run metadata (#396)

• Top-level run APIs accept bounded metadata that is propagated into results, traces, and checkpoint restores.

Zero-key onboarding (#414)

• create-oma-app@0.5.0 can install and run a deterministic Demo without API keys or model requests.
• Demo output clearly discloses simulated model responses and produces Markdown, JSON, and HTML reports.
• New --no-install and --no-run flags support controlled scaffolding workflows.

Examples catalog (#412)

• Added a machine-readable examples catalog, schema, and coverage validation.

Fixes (#390, #417)

• The process backend cleans up descendant processes after the parent exits.
• v1.12.1 fixes installed oma binaries so the CLI launches correctly through npm-created symlinks. This was found during the v1.12.0 post-publish verification.

Compatibility

• Existing run modes remain compatible.
• Evaluation is disabled by default, and online evaluation does not change the business result.
• Scorer failures are recorded as scorer_error and excluded from score aggregates instead of being counted as zero.
• @open-multi-agent/otel@0.1.0 is not republished and remains compatible with core 1.12.1.
• #397, #400, #401, and #402 are behavior-preserving internal extractions, not new public APIs.

Install

npm install @open-multi-agent/core@1.12.1
npm create oma-app@latest my-oma

1.11.0 - 2026-07-17

Features

• Observability v2: trace spans and run identity (#371, #373 by @JackChen-me). Every run now carries a stable runId, attempt, traceId, and rootSpanId, and emits the new TraceRecord v2 schema: a proper span tree covering run, coordinator, task, agent, LLM, tool, retry, delegation, consensus, and checkpoint, with DAG, synthesis, and restore relationships expressed as links. The existing seven-field onTrace callback keeps working unchanged. See docs/observability.md.

• Observability v2: sink and exporter lifecycle (#374 by @JackChen-me). The new @open-multi-agent/core/observability subpath ships the public TraceSink / TraceExporter contract plus BatchingTraceSink (bounded queue, batched export, backoff retries, priority-aware drop), CompositeSink, FilteringSink, SensitiveDataProcessor, and LegacyCallbackTraceSink. Tracing stays metadata-only by default: prompts, completions, tool payloads, credentials, and reasoning content are never captured.

• Observability v2: trace stores (#375, #384 by @JackChen-me). A storage-independent TraceStore contract (append, get, query, delete, retention) with two references: the zero-dependency InMemoryTraceStore, and FileTraceStore behind @open-multi-agent/core/observability/file, an append-only NDJSON store with crash-safe compaction and explicit flush, close, and diagnostics. FileTraceStore targets local development, tests, CLIs, and modest single-process services.

• New package: @open-multi-agent/otel 0.1.0 (#376, #385 by @JackChen-me). A separately installable adapter that maps TraceRecord v2 spans onto OpenTelemetry. Its only peer dependency is @opentelemetry/api ^1.9.0 (npm 7+ installs it automatically), it never initializes or replaces the global TracerProvider, and it versions independently of core. It requires core 1.11.0; core 1.10.0 does not contain the v2 APIs. Core itself remains installable and runnable without any OpenTelemetry packages. See docs/observability-migration.md.

• Per-call tool gate (#377 by @scarab-systems). Optional onToolCall on AgentConfig / OrchestratorConfig runs after Zod validation and before execution, returning allow or deny per call. A deny becomes a normal error ToolResult (never a throw), and a throwing or invalid gate fails closed. Documented with a runnable risk-gated bash example in #383. See docs/tool-configuration.md.

• Generic process backend (#378 by @scarab-systems). The new @open-multi-agent/core/process subpath runs a local command as an agent: protocol-neutral, with stdin or argument prompt delivery, cancellation, and redacted stderr on failure. It complements the existing protocol-aware ACP backend. See docs/external-agents.md.

• Bash command risk classifier (#382 by @JackChen-me). classifyBashCommand behind the @open-multi-agent/core/classifiers subpath returns a safe / review / high level with a reason, splitting compound commands and taking the highest risk. It is a dependency-free heuristic meant to pair with the tool gate, explicitly not a security boundary.

• create-oma-app 0.4.0: production starters (#370 by @JackChen-me). npm create oma-app@latest now offers production-ready pr-review and security starters alongside demo, with cloud or Ollama provider selection. Starters generate Markdown, JSON, and OMA DAG dashboard reports, analyze GitHub strictly read-only, and redact secrets before model analysis.

• Adaptive customer support recipe (#386 by @JackChen-me). A new cookbook example where runTeam engages only the specialists relevant to each ticket instead of the full roster.

Fixes

• Cancelled and timed-out runs no longer report success (#371 by @JackChen-me). Cancellation and whole-run timeout paths could previously surface as successful results; they now report their real outcome through the normalized status.

Compatibility notes

• onTrace remains fully supported and is not deprecated. LegacyCallbackTraceSink can now be configured directly as a sink, preserving the seven legacy event shapes without also wiring onTrace.
• Results now always include identity and status. The new fields stay optional in TypeScript for this release, and success is still present, derived from status.code === 'ok'.
• Dependency ownership is now expressed per package: OpenTelemetry packages belong exclusively to @open-multi-agent/otel, and core must stay importable without them.

Install

npm install @open-multi-agent/core@1.11.0
npm install @open-multi-agent/otel@0.1.0   # optional OpenTelemetry adapter, requires core 1.11.0

New project: npm create oma-app@latest (scaffold updated to create-oma-app@0.4.0, which pins core 1.11.0).

Thanks to @scarab-systems and @JackChen-me.

1.10.0 - 2026-07-11

Features

• External coding agents over ACP (#360 by @JackChen-me). OMA can now orchestrate an external coding agent over the Agent Client Protocol as a first-class team member, alongside LLM agents in the same task DAG. Declare an agent with backend: { kind: 'acp', ... } and it runs a local coding CLI (for example Claude Code via the official @agentclientprotocol/claude-agent-acp adapter) instead of an LLM, sharing the same shared memory, cascade-on-failure, and token budget as any other agent. See docs/external-agents.md.

• Orchestrator cost budget (#358 by @LambIessz). New maxCostBudget and estimateCost on orchestrator config let you set a user-defined spend cap. The estimator receives the effective model, provider, phase, and task id so you can price per model, and the cap is enforced across runAgent, runTasks, runTeam, consensus, and synthesis, mirroring the existing token-budget boundaries.

• Per-agent scoped tool credentials (#362 by @JackChen-me). New AgentConfig.credentials is a per-agent secret bag threaded into ToolUseContext.credentials. Tool code reads context.credentials?.SEARCH_API_KEY instead of closing a single shared secret over every tool, so each agent holds only the credentials it was assigned. A compromised or misbehaving subagent no longer inherits the coordinator's full access.

• Secret redaction on the persistence path (#359 by @JackChen-me). Redaction previously stopped at the observability layer, so a secret an agent emitted into its answer was scrubbed in traces but written in the clear to shared memory and any on-disk checkpoint. The new RedactingStore, a MemoryStore decorator, redacts values at the single choke point every shared-memory and checkpoint write passes through, and is structure-aware for JSON so checkpoint snapshots stay valid with only secrets masked. Closes #339.

• MessageBus persisted in checkpoints (#363 by @LambIessz). Checkpoint and restore now carry MessageBus state (messages and per-agent read position), so a resumed run keeps its inter-agent message history. Live callback subscribers stay process-local and are not serialized. Closes #343.

• Trace span parent linkage (#354 by @tlysanhuo). Every trace event now carries a spanId and optional parentId, with stable agent span ids so LLM and tool spans point to their owning agent span, worker agents link under their task span, and plan_ready links under the coordinator decomposition. A team run is now reconstructable as a proper span tree. Closes #340.

Fixes

• Windows: kill the full process tree on bash timeout (#357 by @Bobuyoucrypto). A timed-out bash command on Windows now kills its full process tree with taskkill /T /F and returns promptly, instead of leaving background children alive and hanging until they exit on their own. Exit codes 124 (timeout) and 130 (abort) are reported authoritatively, and the suite now passes on Windows.

Install

npm install @open-multi-agent/core@1.10.0

New project: npm create oma-app@latest (scaffold updated to create-oma-app@0.3.1, which pins core 1.10.0).

Thanks to @LambIessz, @tlysanhuo, @Bobuyoucrypto, and @JackChen-me.

1.9.0 - 2026-07-03

Features

• Durable memory: FileStore (#347 by @JackChen-me). A zero-dependency, filesystem-backed MemoryStore, so checkpoint/resume survives a process restart out of the box. Writes are atomic (temp file, fsync, rename) and reads come from an in-memory mirror; a corrupt state file fails loud instead of silently. Until now the only bundled store was InMemoryStore, so durability meant writing your own.
• Error-aware task retry (#346 by @JackChen-me). Retry (opt-in via maxRetries > 0) now classifies failures: isRetryableError() skips provably-terminal errors (most 4xx, token-budget, aborted calls) instead of burning attempts, while 429, 5xx, network blips, and per-call timeouts still retry. Adds equal jitter and honors abort signals, on both streaming and non-streaming paths.
• Per-call LLM timeout (#344 by @JackChen-me). New opt-in AgentConfig.callTimeoutMs (and on CoordinatorConfig) bounds a single adapter.chat() call, so one stalled request no longer hangs the whole run. Uniform across every adapter and surfaced as LLMCallTimeoutError, distinct from a deliberate abort or a real API error. Unset preserves current behavior exactly.
• Run-level metrics on TeamRunResult (#345 by @lesbass). TeamRunResult now carries a metrics rollup (total tokens, retries, error/failure/completed counts, and task-latency aggregates) computed from per-task data, and the dashboard renders a matching summary bar.
• Vercel AI SDK 7 support (#348 by @JackChen-me). The optional ai peer range now spans ^5 || ^6 || ^7, so you can use the AI SDK bridge on the latest release without peer-dependency warnings. No adapter code changes. The AI SDK 7 bridge needs Node >= 22 (core stays >= 18).

Docs

• Plan preview and replay guide, surfaced from both READMEs (#349 by @JackChen-me).

Install

npm install @open-multi-agent/core@1.9.0

New project: npm create oma-app@latest

Thanks to @lesbass and @JackChen-me.

1.8.1 - 2026-06-27

Patch release with two install/runtime fixes.

Fixes

• defaultModel now reaches every agent, not just the coordinator. Previously OrchestratorConfig.defaultModel only applied to the coordinator pass. Workers, runAgent, delegated, and consensus agents never inherited it, so it was effectively dead for executing agents. They now inherit defaultModel the same way they already inherited defaultProvider, defaultBaseURL, and defaultApiKey. AgentConfig.model becomes optional in an orchestrated run; a standalone new Agent() still requires an explicit model and throws a clear error if it is missing. (#323)
• Vercel AI SDK v6 is now supported. The optional ai peer range widened from ^5.0.0 to ^5.0.0 || ^6.0.0, so installs that pair the framework with AI SDK v6 (for example via @ai-sdk/react@3, which pulls ai@6) no longer fail with ERESOLVE. The adapter works against both majors. (#324)

Docs

README and examples cleanups: clearer out-of-the-box vs peer-install provider guidance, a one-click Vercel deploy starter link, corrected example run paths after the reorg, and Ecosystem additions. (#320–#322, #325–#331)

---

Also republishes create-oma-app@0.2.1 (template pin bumped to core 1.8.1).

1.8.0 - 2026-06-19

Features

• Checkpoint and resume (#294 by @mvanhorn, #314 by @JackChen-me). A long run can now survive a crash or restart. Persist progress to any MemoryStore with checkpoint: true on runTeam / runTasks / runFromPlan, then resume with orchestrator.restore(...). On resume, runTeam re-runs coordinator synthesis so you still get a final answer, and checkpointing into the team's own shared-memory store no longer re-serializes the whole store per task.
• Consensus verification reaches runTeam (#301 by @nuthalapativarun). The per-task judge loop from 1.7.0 now applies to coordinator-generated tasks. Pass the judges with RunTeamOptions.verifyJudges, and the coordinator opts a task in with verify: true or a partial config.
• Native reasoning round-trip on Bedrock (#302 by @nuthalapativarun). Extended-thinking blocks round-trip through Bedrock's signature protocol with full fidelity, moving echoesReasoning from never to own-issued.

Fixes

• Dashboard: the run details panel is back to a right-hand sidebar on desktop instead of stacking under the canvas (#308 by @JackChen-me).

Onboarding and examples

• npm create oma-app scaffolds a runnable multi-agent demo in one command (#305, #306 by @JackChen-me).
• The full-app examples run end to end on a single DeepSeek key (#311 by @JackChen-me).
• New consensus pattern example (#297 by @nuthalapativarun).

Install

npm install @open-multi-agent/core@1.8.0

New project: npm create oma-app@latest

Thanks to @mvanhorn, @nuthalapativarun, and @JackChen-me.

1.7.0 - 2026-06-15

⚠️ Breaking: built-in tools are now opt-in / default-deny (#289 by @JackChen-me)

A no-tools agent used to receive every built-in implicitly, including an unsandboxed bash. Tool output flows back to the model, so under prompt injection that was a remotely triggerable exec + exfiltration path. Built-in tools (bash, file_*, grep, glob, delegate_to_agent) now need a positive grant: with neither tools nor toolPreset set, an agent resolves to zero.

Migrating:

• One-line restore of the old allow-all: defaultToolPreset: 'full' on OrchestratorConfig.
• Or grant per agent via tools / toolPreset.
• Custom tools (customTools / addTool) are unaffected: registration is the grant, and disallowedTools is still honored.

Features

• Consensus / adversarial verification (#280 by @CodingBangboo). runConsensus() runs proposers against judges, plus an optional per-task verify hook that puts a task's own result through the same judge loop.
• Deterministic model routing (#286 by @cat0825). Opt-in modelRouting sends different orchestration calls (coordinator, synthesis, workers, delegated) to different models by match rules, without mutating your team config. Also fixes routing being bypassed on pooled agents.
• MiniMax-M3 is the new default (#292 by @octo-patch). Up to a 1M-token context (512K guaranteed) with image input. M2.7 and MiniMax-M2.7-highspeed stay available if you pin them explicitly.

Docs

• Model routing guide: docs/model-routing.md, linked from both READMEs (#293 by @JackChen-me).
• TencentDB-Agent-Memory integration cookbook under examples/ (#295 by @JackChen-me).
• LiteLLM added to the OpenAI-compatible providers table (#283 by @RheagalFire).

Install

npm install @open-multi-agent/core@1.7.0

Thanks to @CodingBangboo, @cat0825, @octo-patch, and @RheagalFire. Extra thanks to @CodingBangboo, who landed the consensus primitive (#280) and has been a steady reviewer this cycle.

1.6.0 - 2026-06-06

Features

• Replay persisted team plans, so a previously decomposed plan can be re-run without a new coordinator pass (#285 by @cat0825)
• Add a Tencent Hunyuan provider adapter (#281 by @KaitlynFeng)
• Add a robust regex fallback for malformed single-string JSON tool calls from local models (#269 by @apollo-mg)

Fixes

• Validate message content at every adapter entry, surfacing a clear InvalidMessageError instead of a deep content.some is not a function crash (#288 by @JackChen-me, implementing the fail-fast approach @apollo-mg diagnosed in #268)
• Preserve memoryScope and retry config when a plan is replayed (#287 by @JackChen-me)
• Support structured shared-memory handoff between agents (#284 by @cat0825)

Internal

• Extract a shared repairToolArgs helper across the OpenAI-compatible adapters (#282 by @JackChen-me)
• Add unit tests for the error types (#290 by @Oxygen56)

Docs

• Move the Ecosystem section above Examples in the README (#277 by @JackChen-me)

Install

npm install @open-multi-agent/core@1.6.0

Thanks to @KaitlynFeng, @apollo-mg, @cat0825, and @Oxygen56.

1.5.0 - 2026-05-30

Features

• Per-agent filesystem sandbox (#264 by @JackChen-me). file_read/file_write/file_edit/grep/glob now resolve every path, symlinks included, inside each agent's cwd, defaulting to <cwd>/.agent-workspace. bash stays unsandboxed. Behavior change below.
• Cross-provider reasoning text fallback (#260 by @MyPrototypeWhat). preserveReasoningAsText carries reasoning across providers that can't echo native reasoning blocks, as inline <thinking> text instead of dropping it. Phase 2 of #223.
• MiMo provider (#265 by @kidoom).
• Doubao (Volcengine) provider shortcut (#261 by @kidoom, #236 by @janelawrence). Target Doubao models without custom baseURL wiring.
• DeepSeek V4 default model names (#250 by @JackChen-me). Behavior change below.

Fixes

• Secrets redacted from traces, bash output, and dashboard payloads (#263 by @JackChen-me). API keys and tokens no longer leak into observability surfaces.
• MCP: dedupe normalized tool names, clean up client/transport on connect failure (#256 by @JackChen-me).
• Orchestrator honors abortSignal before synthesis and rejects ambiguous task deps (#255 by @JackChen-me).
• Loop detector replays history and warns on text-only loops (#254 by @JackChen-me).
• DeepSeek echoes reasoning_content on V4 tool-calling (#251 by @JackChen-me).

Behavior changes

• Filesystem tools sandbox to <cwd>/.agent-workspace by default (#264). Agents that read or wrote outside process.cwd(), or used relative paths, now get a sandbox error. Set OrchestratorConfig.defaultCwd (or AgentConfig.cwd) to your root, process.cwd() for the old wide default, or null to disable. bash is not sandboxed.
• DeepSeek CLI default is now deepseek-v4-flash (#250). Affects the oma CLI run without --model; library users are unaffected (no hardcoded default in the adapter). deepseek-chat and deepseek-reasoner keep routing until DeepSeek retires them on 2026-07-24.

Package metadata

• docs/ no longer ships in the npm tarball. README links now point to the GitHub copy, so they always track the latest version. If you read docs from node_modules/@open-multi-agent/core/docs/ (uncommon), switch to the GitHub URLs.

Examples and docs

• Provider examples: Moonshot (#259), Qwen (#257), and an index of existing ones (#258), all by @goodneamtakenbydogs.
• Bilig WorkPaper MCP integration example (#247 by @gregkonush).
• README restructured with built-in capabilities surfaced and zh synced (#273 by @JackChen-me), CLAUDE.md slimmed to a code map (#274 by @JackChen-me), integrations governance README replacing the stale DECISIONS.md (#248 by @JackChen-me).
• Tests and CI: DeepSeek tool-calling via chat() and stream() (#252 by @btroops), OpenAI adapter fallback assertions (#253), CI on npm ci (#262), and a CI check asserting the npm tarball ships only dist/ plus metadata.

Install

npm install @open-multi-agent/core@1.5.0

Thanks to @kidoom, @janelawrence, @MyPrototypeWhat, @goodneamtakenbydogs, @btroops, and @gregkonush.

1.4.2 - 2026-05-24

Drop-in safe patch: three opt-in additions and three new examples. All new APIs default to off.

New (opt-in)

• RunTeamOptions.revealCoordinator (#245 by @JackChen-me). When true, prepends a team-context block (goal, full roster, this worker's assignee identity) to every worker prompt under runTeam. Default false keeps existing prompts byte-identical. runTasks and the short-circuit single-agent path ignore it. Closes #244.
• enableReasoningTextReplay (#234 by @matthewYang08). Opt-in replay of framework reasoning blocks as inline <thinking> text on OpenAI-family requests, capped by maxReasoningReplayChars (default 1200). Also closes a latent 400 against OpenAI when a reasoning-only assistant message emitted {content: null} with no tool_calls. Refs #223.
• ReasoningBlock.provenance + LLMAdapter.capabilities (#243 by @MyPrototypeWhat). Phase 1 of #223: additive IR fields, per-adapter echoesReasoning: 'never' | 'own-issued' declarations, and a shared reasoningBlockToInlineText helper. No behavior change yet; Phase 2 wires capabilities into the outbound fallback.

Examples

• Doubao (ByteDance) provider example via Volcengine Ark (#232 by @nuthalapativarun).
• Zhipu GLM provider example (#231 by @nuthalapativarun).
• Narrative puzzle hint arbitration cookbook: 5-agent multi-source conflict resolution with external safety veto (#235 by @suans4746-del). Closes #212.

Documentation

• compressToolResults and maxToolOutputChars documented in the context management guide (#233 by @nuthalapativarun).

Install

npm install @open-multi-agent/core@1.4.2

Thanks to @MyPrototypeWhat, @nuthalapativarun, @matthewYang08, and @suans4746-del.

Full Changelog: https://github.com/open-multi-agent/open-multi-agent/compare/v1.4.1...v1.4.2

1.4.1 - 2026-05-18

Patch release: two bug fixes for OpenAI-family adapters and the runner, plus Vercel AI SDK as an optional adapter. The previous npm package (@jackchen_me/open-multi-agent) is now formally deprecated.

Bug fixes

• OpenAI-family adapters: guard choices[0] with optional chaining (#220). Prevents crashes when providers return empty choices arrays. Thanks @dvirarad.
• Runner: defer maxTokenBudget break until after tool_result is appended (#221). Fixes orphaned tool_use blocks when the token budget is hit mid-turn. Thanks @CodingBangboo.

New (optional)

• Vercel AI SDK adapter via the dedicated @open-multi-agent/core/ai-sdk subpath (#229). Bridges LLMAdapter to AI SDK's generateText / streamText. Optional peer dep ai; main import is unaffected if you don't use it. Thanks @ibrahimkzmv.

Documentation

• MiniMax provider community offer and setup guide (#219).
• Label hero GIF as post-run replay (#222).

Deprecation

• The previous npm package, @jackchen_me/open-multi-agent, is now formally deprecated. Existing installs continue to work; new installs should use npm install @open-multi-agent/core. The npm CLI shows a deprecation message automatically.

Full Changelog: https://github.com/open-multi-agent/open-multi-agent/compare/v1.4.0...v1.4.1

1.4.0 - 2026-05-09

Highlights

Official org package

Open Multi-Agent now has an official organization package:

npm install @open-multi-agent/core

New projects should use @open-multi-agent/core.

Plan-only orchestration

Adds PlanOnly mode so teams can inspect the coordinator's task DAG before running agent work. (#203 by @CodingBangboo)

LLM adapter improvements

• Preserve reasoning blocks across Anthropic and Gemini turns. (#205 by @MyPrototypeWhat)
• Forward reasoning_effort and backfill sampling-parameter parity across OpenAI-compatible, Copilot, and Azure paths. (#209 by @MyPrototypeWhat)
• Add a Mistral provider example and README entry. (#206 by @mvanhorn)

Shared memory TTL

SharedMemory entries can now expire by turn count. (#213 by @MyPrototypeWhat)

Fixes

• Keep text-tool extraction depth non-negative when a stray closing brace appears. (#217 by @voidborne-d)
• Fix truncation behavior and tighten coordinator dependency guidance. (#215 by @CodingBangboo)

Examples and Docs

• Add paper replication triage cookbook example. (#202 by @DaiMao-UT)
• Add rare disease information triage example. (#211 by @oooooowoooooo)
• Refresh README, hero animation, badges, docs, and repository links for the new GitHub organization. (#214 and #218 by @JackChen-me)

Compatibility

No intentional runtime API breaks were introduced. The package identity changed to @open-multi-agent/core.

The previous package path, @jackchen_me/open-multi-agent, remains supported during the migration window and is also published at 1.4.0.

Install

npm install @open-multi-agent/core@1.4.0

Legacy path during the migration window:

npm install @jackchen_me/open-multi-agent@1.4.0

1.3.1 - 2026-05-02

Features

Streaming reasoning events

StreamEvent now supports a reasoning type that carries the model's thinking tokens in real time. ReasoningBlock is also added to the ContentBlock union for non-streaming paths. Supported on Anthropic and OpenAI providers. (#174 by @SiMinus)

onAgentStream and onPlanReady hooks

Two new orchestrator hooks (runTeam only): onAgentStream delivers real-time per-token streaming events during agent runs, and onPlanReady fires after the coordinator decomposes the goal into a task DAG -- return false to abort before any agent work starts. (#182, #181 by @tizerluo; #184, #183 by @JackChen-me)

Agent Observation Pipeline: new trace events

plan_ready and agent_stream trace events join the trace pipeline, enabling downstream observers to react to plan generation and streaming agent output. (#188 by @ibrahimkzmv)

AWS Bedrock adapter

New LLM adapter for Amazon Bedrock, supporting the full adapter contract (chat + stream). (#194 by @CodingBangboo)

ToolCallTrace includes input/output

ToolCallTrace now carries the tool's input and output payloads, making it useful for debugging and audit without inspecting the raw conversation. (#124 by @MyPrototypeWhat)

Fixes

• Strip image blocks before summarize compression to avoid ballooning token cost. (#196 by @MyPrototypeWhat)
• Preserve tool_use/tool_result pairing during sliding-window truncation, fixing orphaned tool blocks. (#193 by @MyPrototypeWhat)
• onAgentStream path now forwards the full RunOptions into the streaming runner so onTrace, delegation, and run metadata work during streaming. (#184 by @JackChen-me)
• onPlanReady abort path now reports the real coordinator token cost instead of zero, and catches thrown callbacks. (#183 by @JackChen-me)

Examples

• Express customer support pipeline: multi-agent triage, routing, and resolution. (#191 by @CodingBangboo)
• Personalized interview simulator: dynamic question generation with structured output. (#189 by @mmjwxbc)
• Incident postmortem DAG: reconstruct timeline from logs and deploys. (#187 by @binghuaren96)

Docs

• README capability tagline, per-area contributor attribution, and a JSDoc fix. (#190, #186, #180 by @JackChen-me)

Install

npm install @jackchen_me/open-multi-agent@1.3.1

Thanks to @SiMinus, @tizerluo, @ibrahimkzmv, @CodingBangboo, @MyPrototypeWhat, @mmjwxbc, and @binghuaren96 for the external contributions.

1.3.0 - 2026-04-26

New capabilities

Agent delegation

Agents in an orchestrated run can now hand a sub-prompt to another agent on the team and receive its final output as a tool result. Opt-in via registerBuiltInTools(registry, { includeDelegateTool: true }). Five guards: self-delegation, unknown agent, cycle detection, configurable depth cap (maxDelegationDepth, default 3), and pool deadlock. Delegated runs' token usage rolls into the parent's maxTokenBudget so sub-agents cannot silently bypass it. (#123 by @JackChen-me)

runTeam DAG dashboard CLI

oma runTeam ... --dashboard writes a static HTML view of the resolved task graph after a run, including dependencies and per-task status. (#122 by @ibrahimkzmv, follow-up docs in #141 by @JackChen-me)

outputSchema enforcement and defineTool passthrough

The previously advisory outputSchema on AgentConfig is now enforced: results are parsed and validated, with one retry on validation failure. defineTool schemas pass through to the LLM provider. (#149 by @Xin-Mai)

Pluggable shared memory

TeamConfig.sharedMemoryStore accepts any MemoryStore implementation (Redis, SQLite, your own). sharedMemory: true keeps the existing in-process default. (#157 by @JackChen-me)

Advanced LLM sampling

top_p, top_k, repetition_penalty, min_p, and extraBody are now first-class on agent and coordinator configs. Payload spread order is fixed so extraBody overrides sampling parameters but never transport. (#163 by @apollo-mg)

parallelToolCalls exposed for OpenAI

Was previously hardcoded; now configurable per agent. (#173 by @JackChen-me)

Two new providers

• Azure OpenAI adapter, closing the long-standing #24. (#143 by @Klarline)
• Qiniu provider for users on Chinese infrastructure. (#154 by @JackChiang233, follow-up README/CLI docs in #165 by @JackChen-me)

Fixes

• Context compaction persistence and turn dropping. compact strategy was losing turns and not persisting compressed history. (#161 by @apollo-mg)
• OpenAI mixed-content message ordering. Tool messages must precede user messages in mixed content; previously emitted in the wrong order. (#178 by @voidborne-d)
• Provider type widening on configs. AgentConfig, CoordinatorConfig, and OrchestratorConfig were not using the full SupportedProvider union. (#158 by @JackChen-me)

Behavior changes

#163 removed two implicit defaults that some users may have relied on:

• parallel_tool_calls: false is no longer forced. If you need the old behavior, set parallelToolCalls: false explicitly (now exposed via #173).
• The default frequency_penalty override has been removed.

These are behavior changes, not API breaks, but worth checking if you depended on the old defaults.

The same PR also moved the local <think> tag parsing out of the agent layer into tool/text-tool-extractor.ts. This is internal cleanup with no user-visible impact.

Examples and cookbook

Nine new examples and a category reorganization (#125 by @JackChen-me):

• Meeting summarizer pattern. (#139 by @mvanhorn, moved into cookbook/ in #140 by @JackChen-me)
• Translation / back-translation cookbook. (#145 by @zouhh22333-beep)
• Competitive monitoring. (#146 by @pei-pei45)
• Multi-perspective code review, upgraded to structured output and free providers. (#150 by @Kinoo0)
• Contract review DAG with step-level retry. (#155 by @fault-segment)
• Research aggregation with schema. (#159 by @Optimisttt)
• Engram integration: memory store, toolkit, two demos. (#160 by @Agentscreator, Ecosystem section refresh in #151 by @JackChen-me)
• @agentsonar/oma integration: sidecar from the agentsonar team detecting cross-run delegation cycles, repetition, and rate bursts. (e7aecf3 by @JackChen-me)
• Cost-tiered pipeline comparing flagship vs mixed model tiers. (#164 by @HuXiangyu123)
• OpenRouter provider example. (#167 by @kenrogers)
• local-quantized.ts showing tuned sampling on vLLM and llama-server. (ff987cf by @JackChen-me)

Docs and infrastructure

• README refresh: positioning, branding, hero block, integrations, examples section. (#126, #176, #177, #179 by @JackChen-me)
• CLAUDE.md architecture map synced with the current src/ layout. (#171 by @JackChen-me)
• CLI dashboard documented and added to the flag table. (#141 by @JackChen-me)
• Real badges and Codecov integration. (#127, #128, #129, #130 by @JackChen-me)
• npm registry pinned to npmjs.org via repo-level .npmrc (#170 by @JackChen-me).
• Extended LLM adapter coverage for issue #54. (#144 by @jadegold55)

Install

npm install @jackchen_me/open-multi-agent@1.3.0

Thanks to @ibrahimkzmv, @mvanhorn, @Klarline, @jadegold55, @zouhh22333-beep, @Kinoo0, @apollo-mg, @Optimisttt, @Agentscreator, @pei-pei45, @fault-segment, @Xin-Mai, @HuXiangyu123, @JackChiang233, @kenrogers, and @voidborne-d for the external contributions that make this release.

Full changelog: https://github.com/JackChen-me/open-multi-agent/compare/v1.2.0...v1.3.0

1.2.0 - 2026-04-18

First minor release since 1.1.0. MCP integration, three new LLM providers, context management strategies, a CLI, tool output cost controls, and fixes for abort and error propagation.

Features

• MCP integration. New connectMCPTools() wires any MCP server (stdio) directly into agent tool use. @modelcontextprotocol/sdk is an optional peer dependency. Runnable example at examples/16-mcp-github.ts. (#89, by @ibrahimkzmv)

• Three new LLM providers. First-class provider: 'deepseek' (deepseek-chat, deepseek-reasoner), provider: 'minimax' (global and China endpoints via MINIMAX_BASE_URL), and verified Groq via OpenAI-compatible baseURL in examples/19-groq.ts. (#113 and #114 by @hkalex; #121 by @mvanhorn)

• Context management strategies. New AgentConfig.contextStrategy keeps long runs under token ceilings with four strategies: sliding-window, summarize, compact (rule-based, no extra LLM call), and custom. (#88 by @ibrahimkzmv; #111, #119 by @JackChen-me)

• Tool output cost controls. New AgentConfig.maxToolOutputChars and per-tool ToolDefinition.maxOutputChars truncate large outputs (head + tail with a marker). New AgentConfig.compressToolResults compresses older tool results once the agent has moved on; errors are never compressed. (#110, #115, #116, #117, #118 by @JackChen-me)

• CLI (oma). New binary for shell and CI with oma run, oma task, oma provider, JSON-first output, and stable exit codes. Docs at docs/cli.md. (#107 by @ibrahimkzmv)

• AgentConfig.customTools. Inject tool definitions at config time from the orchestrator. Bypasses preset/allowlist filtering but still respects disallowedTools. (#109, #112 by @JackChen-me)

• glob built-in tool. Find files by glob pattern, sorted by modification time. (#102 by @ibrahimkzmv)

Fixes

• AbortSignal propagation. Abort now reaches tool execution, the Gemini adapter, and the abort queue path. (#104 fixes #99, #100, #101, by @JackChen-me)

• Error event propagation. AgentRunner.run() now surfaces error events to callers. (#103 fixes #98, by @JackChen-me)

Examples

• examples/16-mcp-github.ts: full MCP wiring
• examples/17-minimax.ts, examples/18-deepseek.ts, examples/19-groq.ts: provider quickstarts
• examples/with-vercel-ai-sdk/: Next.js + OMA runTeam() + AI SDK useChat

Docs

• READMEs (EN/ZH) expanded: CLI, MCP, context strategies, tool output control, customTools. ZH caught up with EN on items that shipped in 1.1.

Install

npm install @jackchen_me/open-multi-agent@1.2.0

Thanks to @hkalex, @ibrahimkzmv, and @mvanhorn for the external contributions that make this release.

Full changelog: https://github.com/JackChen-me/open-multi-agent/compare/v1.1.0...v1.2.0

1.1.0 - 2026-04-11

First minor release since 1.0.1. Six new features, two fixes, two new examples, and one behavior change you should read before upgrading.

⚠️ Behavior change (read this before upgrading)

Agents now run with default-deny, dependency-scoped context (#87). An agent only sees results from tasks it explicitly dependsOn, instead of every prior task in the run. This prevents context leakage between unrelated agents and keeps token usage predictable in larger teams.

If your existing teams relied on agents implicitly seeing all prior task output, add explicit dependsOn edges in your task graph. No API change is required for runTeam() users whose coordinator already produces a sensible DAG.

This change was prompted by a combination of competitive analysis (XCLI scopes sub-agent context to a minimum file set + tool allowlist by default) and a public post on X by guk2472 flagging inter-agent context pollution as the real production killer in multi-agent systems. Thanks for the signal.

Features

• AbortSignal support for runTeam() and runTasks() (#69). Cancel a run mid-flight from the caller.
• Skip coordinator for simple goals in runTeam() (#70). Single-agent goals no longer pay the coordinator round-trip.
• Token budget management at agent and orchestrator level (#71). Stops runs that exceed a configured budget instead of silently burning tokens.
• Tool allowlist / denylist / preset (#83). Restrict which tools an agent can call without rebuilding the registry.
• Customizable coordinator (#85). Override the coordinator's model, system prompt, tools, toolPreset, and disallowedTools via CoordinatorConfig.
• Dependency-scoped agent context (#87). See behavior change above.

Fixes

• Per-agent mutex prevents concurrent runs on the same Agent instance from corrupting state (#77).
• Duplicate progress events in the short-circuit path for runTeam() are gone, and completedTaskCount is no longer double-incremented (#82).

Examples

• Multi-source research aggregation (#79)
• Multi-perspective code review (#80)

Docs

• README top fold rewritten and Examples section trimmed (#95)
• Coverage badge updated to 88% (#57)
• DECISIONS.md restructured to signal openness on MCP and A2A

Install

npm install @jackchen_me/open-multi-agent@1.1.0

1.0.1 - 2026-04-05

The 1.0.1, 0.2.0, and 0.1.0 entries are summarized from npm package metadata and Git history because no GitHub Release notes were published for them.

Changed

• Limited the npm package to compiled output and package documentation instead of publishing repository source, tests, examples, and project metadata.
• Added contract tests for the Anthropic, OpenAI, Gemini, and GitHub Copilot adapters, plus opt-in provider E2E tests, raising reported coverage from 71% to 88% (#56).

Compatibility

• The compiled dist/ output and public runtime API are unchanged from 1.0.0.

1.0.0 - 2026-04-05

What's new since 0.2.0

Features

• Structured output — optional outputSchema (Zod) on any agent, with auto-retry on validation failure (#36, #38)
• Task retry with exponential backoff — maxRetries, retryDelayMs, retryBackoff per task (#37)
• Observability — onTrace callback emits structured spans for LLM calls, tool calls, tasks, and agent runs (#40)
• Lifecycle hooks — beforeRun / afterRun on AgentConfig for prompt rewriting and result post-processing (#45)
• Human-in-the-loop — onApproval callback between task execution rounds to gate the next batch (#46)
• Loop detection — detects stuck agents repeating the same tool calls or text, with configurable warn / terminate / custom handler (#49)
• Grok (xAI) adapter — first-class support with dedicated GrokAdapter (#44)
• Fallback tool-call extraction — local models that emit tool calls as plain text are now handled automatically (#47)

Testing & quality

• 340 tests, 71% line coverage across src/ (#53)
• Coverage badge added to README (#55)

Full changelog

https://github.com/JackChen-me/open-multi-agent/compare/v0.2.0...v1.0.0

0.2.0 - 2026-04-03

Added

• Added GitHub Copilot as a first-class LLM provider (#9).
• Added baseURL and apiKey overrides for OpenAI-compatible APIs and local servers such as Ollama, vLLM, and LM Studio.
• Agents can declare an outputSchema. Final JSON output is validated with Zod, exposed through result.structured, and retried once with validation feedback when necessary (#36, #38).
• Added task-level retry with capped exponential backoff and accumulated token usage across attempts (#37).

Fixed

• Blocked tasks are unblocked after their dependencies complete instead of remaining stuck indefinitely.
• randomUUID is imported from node:crypto so the package works on its declared Node.js 18 minimum.

0.1.0 - 2026-04-01

Initial npm release.

Added

• OpenMultiAgent with runAgent() for one-shot work, runTeam() for coordinator-planned teams, and runTasks() for explicit task pipelines.
• Role-based teams with shared memory, inter-agent messaging, dependency-aware scheduling, and parallel execution.
• Anthropic and OpenAI adapters, a public custom-adapter contract, and a typed tool framework with five built-in tools.