Install & Run
npm:
npx @desplega.ai/agent-swarm@2.13.5
# or install globally
npm i -g @desplega.ai/agent-swarm@2.13.5bun:
bunx @desplega.ai/agent-swarm@2.13.5
# or install globally
bun i -g @desplega.ai/agent-swarm@2.13.5Docker:
# Server
docker pull ghcr.io/desplega-ai/agent-swarm:2.13.5
# Worker
docker pull ghcr.io/desplega-ai/agent-swarm-worker:2.13.5
# Dashboard
docker pull ghcr.io/desplega-ai/agent-swarm-ui:2.13.5E2B templates:
# API
agent-swarm-api-2-13-5
agent-swarm-api-latest
# Worker / lead runtime
agent-swarm-worker-2-13-5
agent-swarm-worker-latestagent-fs
Ships agent-fs 0.18.33 (bumped from 0.18.32).
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(skills): systemDefault skill seed patches (2026-11-17) by @desplega-bot in #3235
- docs(guides): show illustrative sources and a named owner in the contractor credential example by @desplega-bot in #3236
- docs(specs): sync TLA+ specs with main (heartbeat, workflows) by @desplega-bot in #3237
- chore(docker): bump claude-code, anthropic sdk, amp, grok, kiro (2026-11-17) by @desplega-bot in #3239
- feat(ui): add an admin "Add me to ranking" button on Home by @desplega-bot in #3238
- fix(scripts): support memoryId alias in memory_rate by @ZedingZhang in #3232
- feat(credentials): fresh usage telemetry and lowest-usage key selection by @capchase-bot in #3240
- feat(ui): show quota pace line with shortage/overage on usage cards by @desplega-bot in #3243
- fix(ui): put the sessions sidebar headers and list footers on the row columns by @tarasyarema in #3242
- chore(deps): bump next from 16.4.1 to 16.4.5 in /apps/templates-ui in the npm_and_yarn group across 1 directory by @dependabot[bot] in #3234
- docs(prompts): route trackers to sheets and diagrams to boards by @desplega-bot in #3241
- perf(apps): run app-sync passes up to 3 at a time by @desplega-bot in #3244
- feat(github): mint installation tokens per repository by @spanishkangaroo in #3182
- docs: link worked-example sources in supplier confirmation guide by @desplega-bot in #3245
- docs(guides): lay out the competitor change brief diagram top to bottom by @desplega-bot in #3246
- test(codex): remove duplicate model contract checks by @tarasyarema in #3248
- fix(ui): tighten the sessions phone screens by @tarasyarema in #3247
- docs(testing): define a contract-based test-line budget by @tarasyarema in #3254
- test(drain): consolidate handoff checks and remove redundant waits by @tarasyarema in #3253
- test(ui): avoid OAuth cleanup race in redirect coverage by @tarasyarema in #3255
- test(worktrees): canonicalize reaper fixture paths by @tarasyarema in #3257
- test(workflows): consolidate retry contracts and isolate fixtures by @tarasyarema in #3258
- test(sessions): replace copied HTTP coverage with real contracts by @tarasyarema in #3259
- fix(assets): let the dashboard's logged-in user write its own personal keys by @desplega-bot in #3261
- fix(ui): keep the phone session screen on the page top by @tarasyarema in #3263
- test(scrubber): skip the pass 5 timing budget in the nightly coverage run by @desplega-bot in #3265
- test(scripts): await QuickJS pool readiness before measuring capacity by @tarasyarema in #3266
- ci: shorten the Merge Gate verdict and cut repeated CI work by @tarasyarema in #3252
- feat(defer): wake on checks done by @desplega-bot in #3260
- fix(ui): size the person's message bubble to the chat column by @tarasyarema in #3268
- docs(testing): preserve the cleanup campaign ledger by @tarasyarema in #3270
- test(models): strengthen catalog assertions and remove duplication by @tarasyarema in #3271
- test(extensions): remove redundant example suites by @tarasyarema in #3273
- test(otel): remove mocked no-op and repeated name coverage by @tarasyarema in #3275
- test(apps): consolidate migration checks through production patching by @tarasyarema in #3272
- test(memory): consolidate storage contracts and repair weak guards by @tarasyarema in #3276
- test(db-client): zero the busy spin in the runTimed backoff test by @desplega-bot in #3277
- test(slack): await socket readiness after config changes by @tarasyarema in #3278
- docs(playbooks): add Goals playbook by @desplega-bot in #3279
- test(codex): consolidate cancellation contracts by @tarasyarema in #3281
- docs(tests): reconcile cleanup evidence and remaining scope by @tarasyarema in #3284
- test(mcp): remove repeated enabled-tool registrations by @tarasyarema in #3285
- test(codex): replace copied pool selection with resolver checks by @tarasyarema in #3286
- fix(sandbox): block shell startup files before applying limits by @tarasyarema in #3269
- chore(docker): bump amp, qwen (2026-11-17) by @desplega-bot in #3280
- fix(ui): recover from malformed saved connection configuration by @ZedingZhang in #3283
- perf(pre-push): run only the nearest affected tests and lint changed files by @desplega-bot in #3287
- docs: record cleanup merges and remaining timing failure by @tarasyarema in #3290
- test(scripts): remove fabricated conformance and duplicate classifiers by @tarasyarema in #3291
- docs: describe the fast pre-push hook and CI as the authority by @desplega-bot in #3292
- test(db): replace fixed timing ceiling with accounting guards by @tarasyarema in #3293
- test(heartbeat): remove obsolete checklist prompt fixtures by @tarasyarema in #3294
- test(slack): consolidate repeated rendering contracts by @tarasyarema in #3274
- docs: track test cleanup coverage and measured reductions by @tarasyarema in #3295
- fix(ui): fold the lead's cut reply and its resume into one answer by @tarasyarema in #3296
- feat(users): add user avatars and an agent-fs file picker by @desplega-bot in #3288
- fix(ui): drop the person's bubble minimum to 8rem by @tarasyarema in #3298
- test: consolidate approval card duplicate guards by @tarasyarema in #3299
- test(memory): keep the purge-yield hook off db-client's lock-holding yield by @desplega-bot in #3300
- test: exercise production approval HTTP handlers by @tarasyarema in #3301
- docs: record approval cleanup and coverage checkpoint by @tarasyarema in #3302
- test: remove copied approval task fallback by @tarasyarema in #3304
- test: repair actionable approval executor fixtures by @tarasyarema in #3305
- test: strengthen approval index migration guard by @tarasyarema in #3306
- test: exercise approval follow-ups through HTTP by @tarasyarema in #3307
- docs: record approval cleanup and coverage checkpoint by @tarasyarema in #3309
- test: remove redundant approval list and expiry cases by @tarasyarema in #3310
- test: consolidate approval dashboard guards by @tarasyarema in #3311
- test: consolidate memory rating behavior guards by @tarasyarema in #3313
- docs: record dashboard and memory cleanup evidence by @tarasyarema in #3314
- fix(ui): open boards and sheets as live embeds in Files and session previews by @desplega-bot in #3289
- test(ui): wait for the lazy task body in the link preview sheet test by @desplega-bot in #3312
- chore(deps): bump azure/setup-helm from 5.0.1 to 5.1.0 by @dependabot[bot] in #3315
- fix: count UTF-8 bytes for workflow step output size by @RakeWangsa in #3267
- test: await approval notification completion by @tarasyarema in #3308
- fix(slack): post Lead follow-up outcomes to the Slack thread by @desplega-bot in #3316
- docs: record scheduling cleanup checkpoint by @tarasyarema in #3319
- feat(tasks): record why a task ended in statusKind by @desplega-bot in #3303
- feat(credentials): opt-in pace ranking for pooled keys by @desplega-bot in #3262
- test(evals): use the subprocess budget for timestamp checks by @tarasyarema in #3320
- docs: record eval and Codex test cleanup evidence by @tarasyarema in #3322
- test(codex): remove duplicate client ID and reset-time checks by @tarasyarema in #3321
- test: consolidate duplicate scheduled-task guards by @tarasyarema in #3318
- test(scheduler): distinguish list filter regressions by @tarasyarema in #3323
- docs: reconcile scheduling and eval cleanup evidence by @tarasyarema in #3325
- test(evals): consolidate scenario registry checks by @tarasyarema in #3324
- test(evals): consolidate task dependency validation by @tarasyarema in #3326
- test(evals): consolidate benchmark mapping checks by @tarasyarema in #3327
- docs: reconcile eval engine audits and merged reductions by @tarasyarema in #3328
- test(evals): consolidate outcome normalization guards by @tarasyarema in #3329
- docs: record eval runner audit and cleanup totals by @tarasyarema in #3331
- test(evals): consolidate scoring examples by @tarasyarema in #3332
- test(evals): check judge errors through the retry handler by @tarasyarema in #3333
- test(evals): distinguish baseline attempt and token inputs by @tarasyarema in #3330
- test(evals): consolidate run summary interval checks by @tarasyarema in #3334
- docs: record eval statistics and summary audit by @tarasyarema in #3335
- test(evals): distinguish summary order and cost fields by @tarasyarema in #3336
- release: v2.13.5 + agent-fs 0.18.33 + docs daily update 2026-11-18 by @desplega-bot in #3353
New Contributors
- @spanishkangaroo made their first contribution in #3182
- @RakeWangsa made their first contribution in #3267
Full Changelog: v2.13.4...v2.13.5