v0.4.0

Summary

Release v0.4.0 is a large feature release. CUGA now supports multi-agent work through a new supervisor runtime with an agent registry and a composition UI. The full CugaAgent graph is retired; CugaLite and CugaSupervisor are the two supported agents. Event-driven agents arrive with channels, triggers and NL→Flow arming, including WhatsApp as a direct channel. The supported container image now includes an experimental memory workspace built on Evolve, with retention controls and a management API. CUGA can also run over the Agent Client Protocol (ACP) via stdio. We welcome first-time contributor @anupamamurthi.

On the agent side, CugaLite adds a pre-execute VERIFY gate for code blocks that change data, and tries alternatives before it declares a task impossible. The final answer is now configurable (LLM instructions, a deterministic function, or both). Policies are isolated per agent, so playbooks no longer apply to other agents. A rejected-call guard stops the agent from repeating identical rejected API calls. The LLM layer adds a watsonx Orchestrate (wxo) platform and preserves WatsonX reasoning. Security work removes internal error details from server and events responses, and Dependabot alerts were addressed. Pull requests now require a DCO sign-off.

Upgrade notes

• The full CugaAgent graph is removed (#581). Use CugaLite or CugaSupervisor.
• The evolve extra now requires altk-evolve>=1.5.2.
• Pull requests now need a DCO Signed-off-by line (#897).

---

What's Changed

Supervisor & multi-agent

• feat(supervisor): multi-agent registry, composition UI, and supervisor runtime by @sami-marreed in #433
• feat: retire full CugaAgent graph; keep CugaLite and CugaSupervisor (#580) by @sami-marreed in #581

Memory (experimental)

• feat(memory): add experimental Evolve management API by @illeatmyhat in #736
• feat(memory): add manual retention controls by @illeatmyhat in #735
• feat(memory): add experimental workspace to the supported image by @illeatmyhat in #737

Events & channels

• feat(events & channels): event-driven agents — channels, triggers, and NL→Flow arming by @anupamamurthi in #602
• feat(events & channels): Events followup tooling, deploy kit, rosters by @anupamamurthi in #603
• feat(events & channels): WhatsApp as a direct channel (Meta Cloud API) by @anupamamurthi in #605

CugaAgent

• feat(cuga-lite): pre-execute VERIFY gate for mutating code blocks by @sami-marreed in #739
• fix(cuga_lite): exhaust alternatives before declaring a task impossible (#704) by @Sergey-Zeltyn in #706

Agent, policy & LLM

• feat(agent): configurable final answer — LLM instructions, a deterministic function, or both (#621) by @segevshlomovIBM in #707
• feat(llm): add wxo platform to LLMManager for watsonx Orchestrate routing by @laredo in #716
• fix(policy): per-agent policy isolation — playbooks no longer bleed across agents (#767) by @offerakrabi in #773
• fix(registry): rejected-call guard — stop re-issuing identical rejected API calls (#599) by @Sergey-Zeltyn in #733
• fix(llm): preserve WatsonX reasoning by @sami-marreed in #796
• fix(llm): correct stale RITS registrations — dead Gemma endpoint and 32k Mistral context window by @Sergey-Zeltyn in #725
• fix(state): stop leaking the host wall clock into frozen-world-clock benchmark prompts (#705) by @Sergey-Zeltyn in #708

Integrations & CLI

• feat(acp): add Agent Client Protocol stdio support by @offerakrabi in #787
• feat(appworld): agent-mode login and remote APIs host; fix reassigned variables by @sami-marreed in #795
• fix(answer): classify compose-and-send AppWorld tasks as ACTION (#728) by @Sergey-Zeltyn in #729
• fix(registry): normalize AppWorld file_system paths to canonical ~/ form (#730) by @Sergey-Zeltyn in #740
• feat(cli): add service instance purge command by @illeatmyhat in #726

Knowledge

• fix(knowledge): lower embedding batch defaults so ingest fits a 4GB pod by @dolev31 in #714
• fix(knowledge): let an explicit env override beat the RAG profile by @dolev31 in #713
• fix(knowledge): ground demo starter in Sovereign Core capabilities by @sami-marreed in #906
• docs: replace Knowledge Agent Sovereign Core PDF by @sami-marreed in #905

Security & dependencies

• fix(security): stop sending internal error details to callers by @haroldship in #681
• chore(security): enforce no exception detail in server responses (#721) by @haroldship in #723
• fix(security): stop leaking exception text in the events layer by @haroldship in #769
• chore(llm): use SHA-256 for the model cache key and record why alert 170 is dismissed by @haroldship in #678
• fix(deps): remediate open Dependabot alerts by @haroldship in #763
• fix(deps): address open Dependabot alerts (anyio, accelerate, undici, adm-zip) by @haroldship in #803

Docs, CI & tooling

• ci: add required DCO check and document the external PR limit by @sami-marreed in #897
• docs(contributing): clarify external contribution policy by @sami-marreed in #802
• fix(skills): update cuga commit skill to not commit single files by @offerakrabi in #749
• fix(tests): parse OAuth authorize URL instead of substring matching by @haroldship in #770

New Contributors

• @anupamamurthi made their first contribution in #602

Full Changelog: v0.3.2...v0.4.0