github desplega-ai/agent-swarm v2.8.2

2 hours ago

Install & Run

npm:

npx @desplega.ai/agent-swarm@2.8.2
# or install globally
npm i -g @desplega.ai/agent-swarm@2.8.2

bun:

bunx @desplega.ai/agent-swarm@2.8.2
# or install globally
bun i -g @desplega.ai/agent-swarm@2.8.2

Docker:

# Server
docker pull ghcr.io/desplega-ai/agent-swarm:2.8.2
# Worker
docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.8.2
# Dashboard
docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.8.2

E2B templates:

# API
agent-swarm-api-2-8-2
agent-swarm-api-latest

# Worker / lead runtime
agent-swarm-worker-2-8-2
agent-swarm-worker-latest

agent-fs

Ships agent-fs 0.18.22 (bumped from 0.18.18).

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

  • perf(workflows): scope definition cleanup to deleted runs by @ZedingZhang in #2740
  • fix(seed): let template updates land on workflows whose only change is a setting by @desplega-bot in #2757
  • docs: fix step replay claim in worker recovery guide by @desplega-bot in #2758
  • docs: lead the AI operating model guide with Agent Swarm's fit by @desplega-bot in #2759
  • test(db-client): drop the wall-clock cap from the busy crowd test by @desplega-bot in #2752
  • fix(trackers): require admin auth to start Linear/Jira OAuth by @desplega-bot in #2756
  • feat(models): fall back to the next model profile on provider errors by @desplega-bot in #2736
  • fix(feedback): reject endpoint URLs that include credentials by @itsgintoki in #2755
  • docs: add a local TEI Compose option to the operator skill by @tarasyarema in #2751
  • fix(chat): prevent splitChatMessage from looping on a surrogate pair at a tiny limit by @Goldloli in #2753
  • fix(comb): treat 0.0.0.0 as unreachable in browser reachability checks by @itsgintoki in #2754
  • docs: add AI operations governance guide for small companies by @desplega-bot in #2742
  • fix: parse YAML block scalar skill descriptions by @desplega-bot in #2763
  • feat(ui): app icons and agent avatars from a link or an agent-fs file by @AsLogd in #2729
  • docs: add small tech team operating model guide by @desplega-bot in #2744
  • chore(deps): bump agent-fs pins to 0.18.22 by @desplega-bot in #2761
  • chore(docker): bump amp (2026-11-05) by @desplega-bot in #2762
  • fix: preserve indented skill metadata and keep chomping by @desplega-bot in #2764
  • feat(ui): agent-fs style markdown on every Streamdown surface by @desplega-bot in #2765
  • fix(ui): markdown snippets share the github Monaco theme by @desplega-bot in #2766
  • fix(ci): deploy when newer main commits skip the docker-and-deploy trigger by @desplega-bot in #2773
  • feat(apps): Chart component, Metric sparkline and a sidebar app shell by @desplega-bot in #2769
  • feat(workflows): saving in the workflow header, a Savings tab and a daily chart by @desplega-bot in #2768
  • feat(savings): recalibrate the session curve from a deployment's own sessions by @desplega-bot in #2770
  • release: v2.8.2 + docs daily update 2026-11-06 by @desplega-bot in #2781

New Contributors

Full Changelog: v2.8.1...v2.8.2

Don't miss a new agent-swarm release

NewReleases is sending notifications on new releases.