Repository Issue Activity (beta)

#749 Document the EveryNToolCalls logical-run contract

#747 EveryNToolCalls counter resets every run, so the reminder never fires across turns

#776 [First-time contributor] Looking for beginner-friendly issues to contribute

#765 system_reminders skip follow-up model calls after tool use within a turn

#713 Feature: persistent cross-session memory for Agency Swarm agents via Dakera

#774 OpenAI-compat: OpenAIChatCompletionsModel plus AsyncOpenAI base_url /v1, not a bare model string

#773 RFC: deterministic guardrail on the tool layer (Grimdall)

#772 Portable regression suite for Agency.get_response() across communication_flows → EvalPort

#742 Cost accounting: per-request long-context tiers and cache-write pricing

#744 MCP OAuth: broken docstring wiring, masked root cause, and silently ignored scopes

#743 Structure-only split of mcp/oauth.py (958) and mcp_manager.py (1093) to meet the 500-line cap

#745 Concurrent runs on one thread silently drop the AfterEveryUserMessage reminder

#691 Small note from HVTracker: would you be open to adding your trust badge?

#719 OpenAI 2.45.0 breaks usage initialization with openai-agents 0.14.8

#723 Code Interpreter integration test does not prove uploaded file execution

#721 Retired Anthropic model breaks live integration tests

#704 Reject path separators in remote file upload filenames

#637 Collaboration opportunity: Connectome — open-source AI.OS for human fulfilment

#658 Stale uv.lock: authlib auth-bypass cluster (1 critical + 3 highs) + fastmcp/mcp OAuth tail; consider wiring Dependabot

#598 Feature request: native advisor/executor consult flow for stronger-model escalation

#600 Add support for MiniMax-M2.7 and Kimi K2 (Moonshot AI) models

#596 Chart Library tool for financial pattern analysis in agent swarms

#594 Proposal: ClawMem adapter for thread persistence across sessions

#595 Enforceable agency chart — cryptographic delegation per agent role in the hierarchy

#593 CI-level governance checks for swarm agent code