v2.7.4
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.7.4
# or install globally
npm i -g @desplega.ai/agent-swarm@2.7.4bun:
bunx @desplega.ai/agent-swarm@2.7.4
# or install globally
bun i -g @desplega.ai/agent-swarm@2.7.4Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.7.4 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.7.4 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.7.4
E2B templates:
# API agent-swarm-api-2-7-4 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-7-4 agent-swarm-worker-latest
agent-fs
Ships agent-fs 0.18.13 (bumped from 0.18.12).
agent-swarm pins agent-fs via the chart image tag and npm packages @desplega.ai/agent-fs and @desplega.ai/agent-fs-just-bash; each agent-fs release needs a chore(deps): bump agent-fs pins PR here.
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat) by @desplega-bot in #2564
- chore(deps): bump agent-browser to 0.41.0 by @desplega-bot in #2565
- feat(comb): copy a file's contents from the header copy button by @desplega-bot in #2555
- fix(tasks): keep the task search backfill converging through a locked database by @desplega-bot in #2556
- chore(docker): bump acp, amp, qwen, opencode, goose (2026-10-28) by @desplega-bot in #2557
- fix(slack): keep table blocks and shared-message files on inbound messages by @desplega-bot in #2561
- fix: resolve literal interpolations in MCP docs by @desplega-bot in #2563
- fix: suppress worker follow-ups on deferred completion by @desplega-bot in #2558
- fix(ui): wait for saved model configs before leaving setup by @ZedingZhang in #2559
- feat(runtime-pools): add boxd as a runtime provider by @MichielMAnalytics in #2354
- feat: support environment variables in workflow script nodes by @TheDonten in #2560
- feat(sessions): groups under the Inbox, no Inbox duplicates, auto-Done by @desplega-bot in #2566
- feat(slack): add SLACK_THREAD_WAKE_ON_ANY_MESSAGE opt-in by @tarasyarema in #2567
- release: v2.7.4 + agent-fs 0.18.13 by @desplega-bot in #2568
New Contributors
- @MichielMAnalytics made their first contribution in #2354
- @TheDonten made their first contribution in #2560
Full Changelog: v2.7.3...v2.7.4
v2.7.3
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.7.3
# or install globally
npm i -g @desplega.ai/agent-swarm@2.7.3bun:
bunx @desplega.ai/agent-swarm@2.7.3
# or install globally
bun i -g @desplega.ai/agent-swarm@2.7.3Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.7.3 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.7.3 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.7.3
E2B templates:
# API agent-swarm-api-2-7-3 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-7-3 agent-swarm-worker-latest
agent-fs
Ships agent-fs 0.18.12 (bumped from 0.18.11).
agent-swarm pins agent-fs via the chart image tag and npm packages @desplega.ai/agent-fs and @desplega.ai/agent-fs-just-bash; each agent-fs release needs a chore(deps): bump agent-fs pins PR here.
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat, workflows) by @desplega-bot in #2541
- chore(docker): bump harnesses (2026-10-28) by @desplega-bot in #2538
- chore(deps): bump agent-browser to 0.40.0 by @desplega-bot in #2540
- perf(ui): load sidebar lists with the tree and refetch less on open by @desplega-bot in #2535
- perf(sessions): store session chain summaries in trigger-kept tables by @desplega-bot in #2534
- fix(ui): prepare the first session before leaving setup by @ZedingZhang in #2542
- fix(ui): use a neutral greeting on Home by @ZedingZhang in #2539
- fix(slack): process an event when its socket ack fails because the client is not ready by @desplega-bot in #2490
- perf(db): index and rewrite the slow queries from the query audit by @desplega-bot in #2531
- feat(deploy): one-click deploy via cloud-init, DigitalOcean image and Railway template by @desplega-bot in #2528
- fix(runner): save the default lead profile on first boot by @ZedingZhang in #2537
- chore: add glama.json for the Glama MCP directory by @desplega-bot in #2546
- feat(evals): run pi against Helmcode direct with a helmcode preset by @desplega-bot in #2548
- chore(deps): bump agent-fs pins to 0.18.12 by @desplega-bot in #2549
- feat(comb): Files turns on by default when a browser can reach agent-fs by @desplega-bot in #2544
- fix(comb): attribute Files-page sends to the comment author and show them in sessions by @desplega-bot in #2545
- chore: show pinned agent-fs version in release notes by @desplega-bot in #2550
- docs(guides): add a dreaming loop guide by @desplega-bot in #2552
- fix(nightly-e2e): gate the Codex chain save on an in-container file check by @desplega-bot in #2554
- feat(agents): custom Slack icon per agent by @marcelpallares in #2551
- perf(tasks): search tasks through an FTS5 token index by @desplega-bot in #2553
- release: v2.7.3 + docs daily update 2026-10-29 by @desplega-bot in #2562
New Contributors
- @marcelpallares made their first contribution in #2551
Full Changelog: v2.7.2...v2.7.3
v2.7.2
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.7.2
# or install globally
npm i -g @desplega.ai/agent-swarm@2.7.2bun:
bunx @desplega.ai/agent-swarm@2.7.2
# or install globally
bun i -g @desplega.ai/agent-swarm@2.7.2Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.7.2 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.7.2 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.7.2
E2B templates:
# API agent-swarm-api-2-7-2 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-7-2 agent-swarm-worker-latest
What's Changed
- fix: skip Slack visuals publish when staging is empty by @desplega-bot in #2488
- docs(playbooks): add the Slack Reply Gate playbook by @desplega-bot in #2492
- feat(evals): publish benchmark as dated snapshots by @desplega-bot in #2491
- feat(seed): seed the dreaming loop with a daily schedule by @desplega-bot in #2443
- fix(onboarding): finish the first task step when its task tree settles by @tarasyarema in #2497
- feat(install): short install URL and an automatic tunnel by @tarasyarema in #2498
- fix(telemetry): keep test and demo swarms out of production telemetry by @tarasyarema in #2499
- ci: bound optional ffmpeg install for Slack visuals by @desplega-bot in #2502
- fix(dreaming): find the previous night on manually triggered runs by @desplega-bot in #2505
- fix(sessions): let every user read schedule-run sessions by @capchase-bot in #2493
- docs(playbooks): document unscored Slack Reply Gate skips and fix decision paging by @desplega-bot in #2495
- chore(pricing): refresh models.dev snapshot for gpt-6.1-terra by @desplega-bot in #2496
- feat(skills): add modernize-pr skill and tighten code-quality review rules by @desplega-bot in #2500
- fix(telemetry): tag the scripts-only compose stack as test traffic by @desplega-bot in #2501
- chore(docker): bump opencode, amp, qwen (2026-10-27) by @desplega-bot in #2504
- fix(codex): persist app-server threads so sub-agent spawns can fork context by @desplega-bot in #2506
- fix(seed-scripts): grade gh-pr-snapshot CI against ruleset required checks by @desplega-bot in #2507
- fix(skills): close modernize-pr gaps found on its first live run by @desplega-bot in #2512
- fix(db): index workflow_run_steps on (runId, status) by @desplega-bot in #2489
- fix(ui-e2e): retry tracker ingest through transient API outages by @desplega-bot in #2509
- fix: declare dreaming workflow gallery metadata by @desplega-bot in #2510
- fix(ui): refetch env presence on mount so a fresh tab sees the org-name gate by @desplega-bot in #2511
- chore(evals): weekly matrix lineup with sonnet and terra at a $25 cap by @desplega-bot in #2513
- feat(ui): rebuild Cmd+K from the route table and a search provider registry by @desplega-bot in #2514
- fix(codex): warn when a login duplicates an existing account by @ZedingZhang in #2508
- docs(readme): onboarding demo as the hero video by @tarasyarema in #2515
- fix: make memory Continue test wait for probe state by @desplega-bot in #2517
- docs(thoughts): onboarding feedback plans, credential validation, video follow-ups by @tarasyarema in #2518
- perf(scripts): reduce native script executor startup overhead (#2008) by @duongynhi000005-oss in #2484
- fix: cap auto-stash entries in worker prompts by @desplega-bot in #2520
- refactor(http): centralize request auth to principal resolution (#1958) by @duongynhi000005-oss in #2503
- chore(deps): bump docker/login-action from 4.6.0 to 4.7.0 by @dependabot[bot] in #2527
- chore(deps): bump nginxinc/nginx-unprivileged from 1.31.6-alpine to 1.31.7-alpine by @dependabot[bot] in #2526
- feat(memory): re-embed stored vectors when the embedding space changes by @desplega-bot in #2487
- fix(e2e): carry the Codex OAuth refresh chain between nightly runs by @desplega-bot in #2516
- fix(scripts): bump-agent-fs also bumps the agent-fs-just-bash adapter by @desplega-bot in #2532
- chore(deps): bump agent-fs pins to 0.18.11 by @desplega-bot in #2529
- fix(ui): match task titles and key Cmd+K search by swarm and user by @desplega-bot in #2519
- feat(onboarding-qa): record mode for the demo takes (telemetry block, ttyd terminal, take.sh) by @tarasyarema in #2530
- release: v2.7.2 + docs daily update 2026-10-28 by @desplega-bot in #2536
Full Changelog: v2.7.1...v2.7.2
v2.7.1
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.7.1
# or install globally
npm i -g @desplega.ai/agent-swarm@2.7.1bun:
bunx @desplega.ai/agent-swarm@2.7.1
# or install globally
bun i -g @desplega.ai/agent-swarm@2.7.1Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.7.1 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.7.1 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.7.1
E2B templates:
# API agent-swarm-api-2-7-1 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-7-1 agent-swarm-worker-latest
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat, workflows) by @desplega-bot in #2483
- fix(setup): switch all agents to the first credential's harness by @tarasyarema in #2460
- fix(fs): return pending invite state and accept link from members/invite by @desplega-bot in #2478
- chore(docker): bump claude-code, cursor, amp, copilot, kiro (2026-10-27) by @desplega-bot in #2481
- fix(runner): preserve pi-mono session cost when long lead session settles or aborts (#2073) by @duongynhi000005-oss in #2480
- fix(prompts): codemode tool-discovery line for codex and opencode by @desplega-bot in #2457
- fix(codex): identify OAuth pool slots in terminal auth errors by @ZedingZhang in #2482
- feat(memory): skip store-progress learnings that near-duplicate a swarm memory by @desplega-bot in #2485
- feat(onboarding): bundle agent-fs and local embeddings, add team invites by @tarasyarema in #2442
- release: v2.7.1 + docs daily update 2026-10-27 by @desplega-bot in #2486
Full Changelog: v2.7.0...v2.7.1
v2.7.0
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.7.0
# or install globally
npm i -g @desplega.ai/agent-swarm@2.7.0bun:
bunx @desplega.ai/agent-swarm@2.7.0
# or install globally
bun i -g @desplega.ai/agent-swarm@2.7.0Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.7.0 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.7.0 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.7.0
E2B templates:
# API agent-swarm-api-2-7-0 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-7-0 agent-swarm-worker-latest
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat, workflows) by @desplega-bot in #2439
- chore(docker): bump amp (2026-10-26) by @desplega-bot in #2438
- chore(deps): bump agent-browser to 0.39.4 by @desplega-bot in #2440
- fix(comb): keyboard-reachable row menus and full menu on Recent and Favorites by @desplega-bot in #2435
- chore(deps): bump agent-fs pins to 0.18.7 by @desplega-bot in #2446
- feat(heartbeat): wake the gate when the script runner is unhealthy by @oscar-izval in #2447
- chore(deps): bump vulnerable docs-site dependencies (Dependabot alerts) by @desplega-bot in #2448
- chore(deps): bump vulnerable root dependencies (Dependabot alerts) by @desplega-bot in #2449
- feat(ui): per-call cache-hit chip in the session log by @desplega-bot in #2450
- feat(workflows): emit workflow.run.finished when any run ends by @desplega-bot in #2452
- feat(pages): let PAGES_FRAME_ANCESTORS origins iframe public pages by @desplega-bot in #2454
- docs: publish lead agent rulebook playbook by @desplega-bot in #2456
- chore(deps): bump agent-fs pins to 0.18.8 by @desplega-bot in #2461
- feat(pi): log one call_usage line per LLM call by @desplega-bot in #2451
- chore(scripts): tool-schema token probe for harness bumps by @desplega-bot in #2455
- feat(ui): request composition panel and per-call prompt bar by @desplega-bot in #2471
- chore(docker): bump dsh, amp (2026-10-26) by @desplega-bot in #2470
- fix(oauth): parse OAUTH_KEEPALIVE_DISABLE and OAUTH_REFRESH_SWEEP_DISABLE via isEnvFlagEnabled by @duongynhi000005-oss in #2467
- fix(workflows): abort executors on step timeout; add trigger-workflow wait:false by @desplega-bot in #2441
- feat(skills): make agents aware of docs.agent-swarm.dev playbooks and guides by @desplega-bot in #2462
- fix(credentials): parse CRED_CHECK_DISABLE via isEnvFlagEnabled by @duongynhi000005-oss in #2468
- feat(scripts): allow selecting a script executor per saved script by @duongynhi000005-oss in #2464
- feat(scripts-runtime): skip duplicate host-validated argsSchema.parse in native executor by @duongynhi000005-oss in #2465
- feat(scripts-runtime): skip duplicate host-validated argsSchema.parse in quickjs executor by @duongynhi000005-oss in #2466
- fix(evals): count only real swarm-docs skill loads in docs-suggestion by @desplega-bot in #2473
- fix(onboard): mount agent volumes where the worker writes by @desplega-bot in #2474
- feat(store-progress): optional learning field writes a swarm memory by @desplega-bot in #2472
- docs: add host requirements and Docker disk cleanup guide by @desplega-bot in #2475
- feat(credentials): match key availability by key suffix, not the global pool index by @desplega-bot in #2436
- docs(playbooks): add the Linear Delivery Loop playbook and template by @desplega-bot in #2463
- chore(deps): bump agent-fs pins to 0.18.9 by @desplega-bot in #2476
- fix(docs): use supported code fence for cron example [preview] by @desplega-bot in #2477
- feat(pi): tool deferral on by default; defer installed MCP servers by @desplega-bot in #2453
- release: v2.7.0 + docs daily update 2026-10-27 by @desplega-bot in #2479
New Contributors
- @duongynhi000005-oss made their first contribution in #2467
Full Changelog: v2.6.0...v2.7.0
v2.6.0
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.6.0
# or install globally
npm i -g @desplega.ai/agent-swarm@2.6.0bun:
bunx @desplega.ai/agent-swarm@2.6.0
# or install globally
bun i -g @desplega.ai/agent-swarm@2.6.0Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.6.0 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.6.0 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.6.0
E2B templates:
# API agent-swarm-api-2-6-0 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-6-0 agent-swarm-worker-latest
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat, workflows) by @desplega-bot in #2413
- chore(evals): publish benchmark 1.0 from weekly run run-202610250200-3822d6 by @desplega-bot in #2414
- chore(docker): bump amp, qwen (2026-10-25) by @desplega-bot in #2411
- chore: re-pin researching skill retrieval guidance by @desplega-bot in #2410
- docs: verify skill writes in script guide by @desplega-bot in #2409
- fix(workflows): break startedAt ties by rowid when reading run steps by @dev-petrov in #2416
- feat(tasks): filter GET /api/tasks by workflowRunId by @dev-petrov in #2417
- feat(db-retention): opt-in retention for finished script_runs by @desplega-bot in #2404
- feat(comb): search, share links, audio and HTML preview, uploads by @tarasyarema in #2406
- fix(onboarding): clean install and /setup flow for new users by @tarasyarema in #2405
- feat(extensions): expose Slack messages in task creation hooks by @ZedingZhang in #2412
- feat(sessions): Slack and chat source sections, source keys, and backfill by @tarasyarema in #2408
- feat(comb): send any comment, bulk resolve, In progress home, Recent and Favorites by @tarasyarema in #2420
- chore(docker): bump amp (2026-10-25) by @desplega-bot in #2421
- fix(sessions): readable thread group labels, the phone filter button, and thread group docs by @tarasyarema in #2423
- fix(credentials): test each pool key, re-test on change, check Claude tokens by @tarasyarema in #2422
- fix(claude): seed trust in the session HOME, not the host ~/.claude.json by @tarasyarema in #2424
- feat(codex): route Codex through a gateway with CODEX_GATEWAY_BASE_URL by @desplega-bot in #2426
- fix(tests): retry the API spawn when its free port is taken by @desplega-bot in #2427
- feat(comb): file menu, filters, search in the URL, agent-fs stars, asset comments by @tarasyarema in #2428
- docs(readme): add the launch video asset by @tarasyarema in #2429
- docs(readme): launch video as the README hero by @tarasyarema in #2431
- feat(workflows): read a run tree via GET /api/workflow-runs/{id}/tree by @dev-petrov in #2419
- fix(workflows): do not retry a sub-workflow step whose child run was deleted by @dev-petrov in #2433
- feat(workflows): add fields=slim to GET /api/workflow-runs/{id} by @dev-petrov in #2418
- feat(credentials): list, add and remove single pool keys, with an instant key check by @tarasyarema in #2430
- docs: Release Notes — Week of October 19 to October 26, 2026 by @desplega-bot in #2434
- feat(skills): onboarding-qa maintainer skill (install, ui and stack modes) by @tarasyarema in #2362
- release: v2.6.0 + docs daily update 2026-10-26 by @desplega-bot in #2437
Full Changelog: v2.5.4...v2.6.0
v2.5.4
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.5.4
# or install globally
npm i -g @desplega.ai/agent-swarm@2.5.4bun:
bunx @desplega.ai/agent-swarm@2.5.4
# or install globally
bun i -g @desplega.ai/agent-swarm@2.5.4Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.5.4 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.5.4 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.5.4
E2B templates:
# API agent-swarm-api-2-5-4 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-5-4 agent-swarm-worker-latest
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat, workflows) by @desplega-bot in #2392
- fix: stop poll claims during worker shutdown by @desplega-bot in #2395
- fix(controller): reconnect after a half-open socket by @desplega-bot in #2396
- fix(ui): stop metadata-only integrations from showing Configured and copy inline env names in Sessions by @desplega-bot in #2394
- chore(docker): bump claude-code, acp, anthropic-sdk, amp, grok, goose, kiro (2026-10-24) by @desplega-bot in #2391
- fix(config): let a script run's token read its held secrets on list and by-id by @desplega-bot in #2393
- fix: remove anonymous volumes when the docker provider deletes a runtime by @MaximilianoAdaro in #2390
- chore(skills): re-sync vendored ai-toolbox skills to a1e33bc by @desplega-bot in #2389
- fix(session-logs): read session logs in write order by @desplega-bot in #2383
- fix(workflows): retry a foreach agent-task item cut by a deploy drain by @desplega-bot in #2370
- docs(tasks): check render and PR state when resuming by @desplega-bot in #2388
- feat(workflows): idempotent create and trigger with an Idempotency-Key header by @dev-petrov in #2380
- chore(deps): bump bun to 1.4.4 by @capchase-bot in #2397
- docs(skills): exempt held reruns with matching CI success by @desplega-bot in #2398
- chore(docker): bump amp (2026-10-24) by @desplega-bot in #2399
- fix(workflows): cancel a workflow run's whole child tree and keep it stopped by @dev-petrov in #2377
- docs: default agent-fs links to the live app link by @desplega-bot in #2400
- feat(extensions): emit post.task.blocked for rejected task creation by @ZedingZhang in #2402
- fix(ui): total the session header cost over the whole task chain by @desplega-bot in #2403
- feat(hooks): one-time context-bound note for long worker sessions by @desplega-bot in #2401
- release: v2.5.4 + docs daily update 2026-10-25 by @desplega-bot in #2407
Full Changelog: v2.5.3...v2.5.4
v2.5.3
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.5.3
# or install globally
npm i -g @desplega.ai/agent-swarm@2.5.3bun:
bunx @desplega.ai/agent-swarm@2.5.3
# or install globally
bun i -g @desplega.ai/agent-swarm@2.5.3Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.5.3 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.5.3 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.5.3
E2B templates:
# API agent-swarm-api-2-5-3 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-5-3 agent-swarm-worker-latest
What's Changed
- feat(skills): rank skill-search results with BM25 by @desplega-bot in #2361
- chore(docker): bump opencode, dsh, amp, anthropic sdk (2026-10-23) by @desplega-bot in #2369
- feat(extensions): comms-adherence-gate judges plain language and agent Slack posts by @desplega-bot in #2360
- fix(ui): keep the border beam a ring in Firefox by @desplega-bot in #2371
- fix(chat-sdk): acknowledge addressed asks in a subscribed Telegram thread right away by @desplega-bot in #2368
- feat(ui): pretty-print raw task prompts in sessions, lists and pop-overs by @desplega-bot in #2373
- fix(workflows): keep one child run per sub-workflow invocation across crashes and retries by @dev-petrov in #2364
- feat(hitl): config question type sets swarm config and secrets safely by @desplega-bot in #2374
- fix(ui): read generated prompt envelopes from their backend builders by @desplega-bot in #2376
- fix(sessions): show only ui sessions by default and keep Done through agent activity by @desplega-bot in #2379
- fix(entrypoint): make the GitLab git credential helper setup idempotent by @dev-petrov in #2372
- docs(skills): ask for missing config and secrets with a config question by @desplega-bot in #2378
- fix(sessions): reopen a Done session when a person steers it by @desplega-bot in #2381
- fix(skills): rank name and description hits first in skill-search by @desplega-bot in #2375
- docs: reserve changelog edits for release PRs by @desplega-bot in #2382
- release: v2.5.3 + docs daily update 2026-10-24 by @desplega-bot in #2387
New Contributors
- @dev-petrov made their first contribution in #2364
Full Changelog: v2.5.2...v2.5.3
v2.5.2
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.5.2
# or install globally
npm i -g @desplega.ai/agent-swarm@2.5.2bun:
bunx @desplega.ai/agent-swarm@2.5.2
# or install globally
bun i -g @desplega.ai/agent-swarm@2.5.2Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.5.2 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.5.2 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.5.2
E2B templates:
# API agent-swarm-api-2-5-2 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-5-2 agent-swarm-worker-latest
What's Changed
- docs(specs): sync TLA+ specs with main (heartbeat) by @desplega-bot in #2355
- chore(deps): bump agent-browser to 0.39.2 by @desplega-bot in #2357
- fix(approvals): post Slack notifications for standalone approval requests by @desplega-bot in #2356
- fix(ci): make check-codex-default-model.sh read the live default and check CODEX_VERSION by @desplega-bot in #2353
- feat(sessions): show only my sessions by default, with a Mine / All toggle by @desplega-bot in #2347
- fix(docker): pin Node 22.23.4 in worker image by @desplega-bot in #2348
- dreaming-v2 systemDefault skill seed patches (2026-10-23) by @desplega-bot in #2351
- chore(docker): bump claude-code, codex, copilot, amp (2026-10-23) by @desplega-bot in #2352
- Add a per-run session token for script runs by @MaximilianoAdaro in #2349
- chore(deps): bump vulnerable apps/templates-ui dependencies (Dependabot alerts) by @desplega-bot in #2358
- chore(deps): bump vulnerable docs-site dependencies (Dependabot alerts) by @desplega-bot in #2359
- feat(rbac): assign RBAC roles to users through the API, add a viewer role by @khoapmdx in #2304
- feat(chat-sdk): Telegram answers reply to their ask, carry a task tag, skip deferrals by @desplega-bot in #2203
- feat(chat-sdk): Telegram reaction lifecycle instead of text acks by @desplega-bot in #2366
- feat(sessions): filter menu, select all, and a one-time Done backfill by @tarasyarema in #2367
- release: v2.5.2 by @desplega-bot in #2365
Full Changelog: v2.5.1...v2.5.2
v2.5.1
Install & Run
npm:
npx @desplega.ai/agent-swarm@2.5.1
# or install globally
npm i -g @desplega.ai/agent-swarm@2.5.1bun:
bunx @desplega.ai/agent-swarm@2.5.1
# or install globally
bun i -g @desplega.ai/agent-swarm@2.5.1Docker:
# Server docker pull ghcr.io/desplega-ai/agent-swarm:2.5.1 # Worker docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.5.1 # Dashboard docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.5.1
E2B templates:
# API agent-swarm-api-2-5-1 agent-swarm-api-latest # Worker / lead runtime agent-swarm-worker-2-5-1 agent-swarm-worker-latest
What's Changed
- fix(gitlab): read the bot name lazily, cancel every MR task on close by @desplega-bot in #2343
- release: v2.5.1 + docs daily update 2026-10-23 by @desplega-bot in #2350
Full Changelog: v2.5.0...v2.5.1