Skip to content

OMTAE brain drift, Qwen chat fixes, B2B MCP, autonomy wiring#1264

Open
jaylee710 wants to merge 9 commits into
RightNow-AI:mainfrom
jaylee710:cursor/omtae-brain-drift-profiles
Open

OMTAE brain drift, Qwen chat fixes, B2B MCP, autonomy wiring#1264
jaylee710 wants to merge 9 commits into
RightNow-AI:mainfrom
jaylee710:cursor/omtae-brain-drift-profiles

Conversation

@jaylee710

Copy link
Copy Markdown

Summary

  • Fix Qwen3.x orphan thinking-tag rendering and empty-session provider errors
  • Add ECC anti-hallucination loop guard and session repair
  • Add B2B revenue MCP server (mcp/omtae-b2b/omtae_mcp.py) for Hermes Desktop
  • Add idempotent scripts/restore-autonomy.sh for desk schedules and hands

Test plan

  • cargo test --workspace (prior session)
  • Live desk health at :4200
  • MCP tools smoke test (discover / analyze / report)
  • Autonomy restore script idempotent run

Made with Cursor

jaberjaber23 and others added 5 commits June 4, 2026 06:00
…file, and drift-guard capabilities so the API and dashboard can manage models and catch config drift reliably.

Co-authored-by: Cursor <cursoragent@cursor.com>
Add loop guard for repeated assistant text, ops-fixer manifest, Qwen3.6 agent defaults, brain API hardening, and model-switch script updates so system checks use real tools instead of hallucinated status.

Co-authored-by: Cursor <cursoragent@cursor.com>
…MTAE agents.

Runtime TOOL_REQUIRED verification, tighter planning-loop detection, tool-first prompts, and bundled ECC skills help desk agents stop narrating without evidence.

Co-authored-by: Cursor <cursoragent@cursor.com>
Strip orphan </think> tags from Qwen3.x streams and inject a
placeholder user turn when repair/trim leaves no user text, so vLLM chat
templates stop failing with "No user query found in messages".

Co-authored-by: Cursor <cursoragent@cursor.com>
Expose niche discovery, site audit, and pitch generation tools for Hermes
Desktop, plus an idempotent script to rewire desk schedules and hands.

Co-authored-by: Cursor <cursoragent@cursor.com>
@jaylee710

Copy link
Copy Markdown
Author

ok

jaberjaber23 and others added 4 commits July 6, 2026 12:24
…ze local inference backend.

Change weekly-brain-pipeline cron schedule to resolve parsing warnings. Support skill execution tools across coder, researcher, and ecc-bridge agents. Optimize and stabilize local vLLM 32k context and gateway auth overrides.

Co-authored-by: Antigravity <antigravity@google.com>
Allow orchestrator agent to run shell commands directly for system automation and diagnostics tasks.

Co-authored-by: Antigravity <antigravity@google.com>
…nd enable wildcard tool capabilities for all agents
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants