feat: multi-modal agent — voice, file sending, vision#644

Closed

brandontan wants to merge 74 commits intoqwibitai:mainfrom

brandontan:feat/multi-modal-agent

Commits on Feb 28, 2026

feat: add Zod-validated LLM output schemas and retrofit observer

brandontan
and
claude
committed
Merge pull request #29 from brandontan/feat/zod-validated-outputs
brandontan
authored
feat: add conversation quality tracker with implicit signal extraction

brandontan
and
claude
committed
Merge pull request #31 from brandontan/feat/quality-tracker
brandontan
authored
feat: add auto-learner — detect corrections and log learnings

brandontan
and
claude
committed
Merge pull request #32 from brandontan/feat/auto-learning
brandontan
authored
feat: add per-step evaluation — validate agent outputs before use

brandontan
and
claude
committed
Merge pull request #33 from brandontan/feat/per-step-eval
brandontan
authored
feat: add reflector — deterministic memory garbage collection

brandontan
and
claude
committed
Merge pull request #34 from brandontan/feat/reflector
brandontan
authored
feat: add structured memory — categorized knowledge storage

brandontan
and
claude
committed
Merge pull request #35 from brandontan/feat/structured-memory
brandontan
authored
feat: add hindsight module — auto post-mortem on frustrated conversations (#5) (#36)

brandontan
and
claude
authored
feat: add workflow router — deterministic config-driven routing (#21) (#37)

brandontan
and
claude
authored
feat: add progressive memory — layered recall with two-stage disclosure (#24) (#38)

brandontan
and
claude
authored
feat: add tool observability layer — structured JSONL logging for all MCP tool calls (#25) (#39)

brandontan
and
claude
authored
feat: add smart model routing — auto-pick optimal model by task type (#7) (#40)

brandontan
and
claude
authored
feat: add task templates — reusable structured prompts with anti-pattern guardrails (#9) (#41)

brandontan
and
claude
authored
feat: add tool guard — pre-execution security layer for all tool calls (#10) (#42)

brandontan
and
claude
authored
feat: tool guardrails audit — rate limits, spend caps, safety framework (#28)

brandontan
and
claude
committed
Merge pull request #43 from brandontan/feat/tool-guardrails-audit
brandontan
authored
feat: self-knowledge — agent can explain its own capabilities (#27)

brandontan
and
claude
committed
Merge pull request #44 from brandontan/feat/self-knowledge
brandontan
authored
feat: ACP adapter — make agents driveable from external tools (#23)

brandontan
and
claude
committed
feat: atomic rollback deploys — symlink-based release management (#11)

brandontan
and
claude
committed
Merge pull request #45 from brandontan/feat/atomic-deploys
brandontan
authored
feat: CLI — sovereign init/deploy/status/logs/rollback (#12)

brandontan
and
claude
committed
Merge pull request #46 from brandontan/feat/cli
brandontan
authored
feat: multi-repo workspaces — per-task git worktrees for delegated work (#13)

brandontan
and
claude
committed
Merge pull request #47 from brandontan/feat/multi-repo-workspaces
brandontan
authored
feat: Slack channel support — Socket Mode, threads, mentions (#14)

brandontan
and
claude
committed
Merge pull request #48 from brandontan/feat/slack-channel
brandontan
authored
feat: sentry agent — automated incident triage via webhooks (#15)

brandontan
and
claude
committed
Merge pull request #49 from brandontan/feat/sentry-agent
brandontan
authored
feat: structured elicitation — numbered options via IPC for agent decisions (#26) (#50)

brandontan
and
claude
authored
feat: agent-to-agent relay messaging — peer-to-peer coordination via IPC (#30) (#51)

brandontan
and
claude
authored
Merge develop into main — v0.4.0 Multi-agent & Channels
brandontan
committed
feat: v1.0.0 Production Ready — clean codebase, plugin system, docs (#52)

brandontan
and
claude
authored

Commits on Mar 1, 2026

Commits on Mar 2, 2026