v0.1.4

Highlights

• publish a stable, versioned container path: ghcr.io/happysnaker/qq-ai-bot:v0.1.4
• keep ghcr.io/happysnaker/qq-ai-bot:latest as the moving mainline tag for people who want the newest default-branch build
• continue shipping the Docker image as a multi-arch manifest for linux/amd64 and linux/arm64
• pin the demo stack docs to the stable v0.1.4 image path so app directories / homelab users can reference a non-moving tag

Why this release exists

This is mainly a packaging / distribution release.

The goal is to make qq-ai-bot easier to reference from self-hosted directories, homelab install notes, and app-store style submissions that prefer a stable container tag instead of only latest.

Stable image paths

• ghcr.io/happysnaker/qq-ai-bot:v0.1.4
• ghcr.io/happysnaker/qq-ai-bot:latest

Notes

• no new runtime feature claims are introduced in this release beyond the stable container-tag packaging work
• multi-arch image publishing remains enabled for both amd64 and arm64

v0.1.3

Highlights

• stage richer-media fallback handling for file / audio / video-like inbound segments
• stop silently dropping unsupported inbound media before the ACP layer sees any hint of them
• attach explicit “未直传媒体” notes to prompts so agents do not pretend they already read unsupported attachments
• document the current richer-media boundary in README, FAQ, roadmap, architecture, and agent integration docs

Validation

• npm run lint
• npm run typecheck
• npm run test

v0.1.2

Highlights

• add per-interaction correlation IDs across receive / dedupe / ACP dispatch / progress / reply logs
• expose duplicate-drop dedupe keys in structured logs for faster replay debugging
• document how operators can grep correlationId to trace a noisy interaction end to end

Validation

• npm run lint
• npm run typecheck
• npm run test

v0.1.1

Highlights

• add lightweight inbound replay dedupe before command handling and ACP dispatch
• expose qq_ai_bot_inbound_duplicates_total in Prometheus-style metrics
• add config knobs for inbound dedupe window and cache size
• sharpen README / configuration docs around replay protection and the next observability contribution path
• close the first replay-dedupe issue and hand off the next engineering task as issue #13

Why this release matters

This patch release pushes qq-ai-bot a bit further away from a happy-path bot demo and a bit closer to message infrastructure:

• duplicate OneBot deliveries are less likely to become duplicate ACP calls
• operators get a runtime signal for replay drops
• the public roadmap now points to correlation IDs as the clearest next production-minded contribution

v0.1.0

Initial public release of qq-ai-bot.

Highlights:

• OneBot 11 forward/reverse WebSocket support
• NapCat / LLOneBot compatibility
• ACP-based AI agent bridge
• session persistence and progress streaming
• Docker-friendly self-hosted deployment

Project page: https://happysnaker.github.io/qq-ai-bot/
Support: https://happysnaker.github.io/support/