CCCC v0.4.9 Release Notes
v0.4.9 builds on v0.4.8 with significant daemon-side reliability work, a new IM bridge, and a broad round of Web and operator-surface improvements.
Compared with v0.4.8, this release unifies actor launch resolution, adds assistive-job truth boundaries for pet and context flows, introduces the WeChat (Weixin) IM bridge, refines presentation workflows, and includes a new text-size accessibility control for the Web UI.
Highlights
1) Actor launch resolution and pet runtime sync were unified
The most impactful daemon change in v0.4.9 is a unified actor launch pipeline.
Actor add, update, lifecycle, and runtime operations now follow a single resolution path with consistent async-result semantics. Group lifecycle start/stop flows share the same actor-status tracking, and the pet runtime gained explicit sync with group settings (including desktop-pet enablement). A new async_result contract formalizes accepted/completed/queued signaling across the daemon IPC surface.
Key improvements:
- actor add/update operations use a shared resolution pipeline with preflight validation
- group start/stop reliably awaits per-actor results instead of fire-and-forget
- pet runtime state tracks desktop-pet group settings and syncs on change
- execution queues enforce stricter ordering guarantees
- MCP dynamic capability tools reflect real-time actor state
2) Truth boundaries were tightened across pet and context flows
v0.4.9 introduces an assistive-jobs layer that enforces canonical readback for task and context writeback paths.
Pet review scheduling was refactored from a monolithic scheduler into distinct job kinds (review, profile refresh) with explicit trigger classification. Context operations now distinguish accepted versus completed status, and pet profile refresh completes only on real assistive outputs rather than fire-and-forget.
Key improvements:
- pet review and profile jobs require verified completion before marking done
- context sync and group-space writeback report accepted vs. completed separately
- immediate pet reviews flush synchronously instead of entering the timer queue
3) WeChat (Weixin) IM bridge
v0.4.9 adds the WeChat bridge as a first-class IM adapter alongside Telegram, Slack, Discord, Feishu/Lark, and DingTalk.
The implementation includes a Node.js sidecar for the WeChat desktop protocol, a CLI login/logout flow, QR-code-based authentication in the Web UI, and daemon routes for bridge lifecycle management. The Weixin adapter follows the same bridge contract and bind-key authorization model as the existing adapters.
4) Presentation workflow and outbound delivery were refined
The presentation viewer gained inline web-preview support, topic-aware slide navigation, and improved full-screen behavior. Outbound delivery through the IM surface now handles richer content types, and the chat tab integrates a split-layout mode for simultaneous conversation and presentation viewing.
5) Web Pet task advisor became local-first
The Web Pet task proposal pipeline was reworked to run a local task advisor that evaluates evidence before surfacing proposals. This reduces noise from speculative suggestions and makes the pet's task-related nudges more reliable. The idle-standup suppression and silence-activity filters also improved, so the pet stays quiet when there is genuinely nothing to act on.
6) Web UI polish and accessibility
A broad round of Web improvements landed in this release:
- Text size control: a new three-tier text-scale selector (90% / 100% / 125%) in the header rail and mobile menu, persisted per-browser via
localStorage. - System theme icon: the system-theme toggle now uses a monitor/display icon instead of the previous terminal icon, improving clarity.
- Mobile theme cycling: the mobile menu now cycles through light → dark → system instead of binary toggling.
- Group sidebar extraction: the sortable group list was extracted into a standalone component with optimized chunk splitting.
- Runtime visibility controls: peer and pet tabs gained explicit show/hide toggles based on actor composition.
- Layout and polling: follow-bottom logic was removed in favor of explicit auto-follow heuristics, and group-space polling behavior was simplified.
- Automation snippet catalog: builtin snippet overrides were separated into a distinct catalog contract for cleaner customization.
Summary
In short, v0.4.9 is the release that brings together:
- a unified actor launch and pet runtime resolution pipeline
- tighter truth boundaries for pet review, context writeback, and group-space operations
- WeChat (Weixin) as a new IM bridge with full login/auth/lifecycle support
- refined presentation workflow with inline preview and split-layout mode
- a local-first pet task advisor that reduces proposal noise
- text-size accessibility, mobile theme cycling, and another round of Web polish