Apply deterministic routing decisions to PRD graph progression after reflect and advance steps.
npx skills add https://github.com/wlee075/chatbot --skill routing-wlee075
Please help me install this Agent Skill. Skill: routing Source: https://github.com/wlee075/chatbot/tree/main/skills/routing Command: npx skills add https://github.com/wlee075/chatbot --skill routing-wlee075
The routing logic provides deterministic decisions for the AI graph flow after reflect and advance_section, ensuring predictable progression through PRD sections and consistent handling of special cases like PASS, recovery caps, and iteration limits.
Feed reflect and advance data into the routing module to obtain a deterministic next action for PRD progression.
High-intent search queries and answers about installing and using this skill.
Deterministic routing decisions for PRD graphs are made by applying a decision table after reflect and advance steps. The routing module computes the next step using state fields like verdict and iteration without mutating state.
Deterministic routing uses a clear priority order: PASS > RECOVERY_CAP > ITER_CAP > LOOP. Each decision emits explicit reason codes and routing_decision logs, ensuring predictable progression through PRD sections.
PRD graph routing handles recovery caps and iteration limits by evaluating recovery_mode_consecutive_count and max_iterations state fields. When caps are hit, the route_after_reflect decision table triggers forced-advance semantics to progress.
Yes, the routing logic supports extensibility by allowing you to add new conditional edges and updated reason codes. This accommodates new routing constraints for deterministic PRD graph progression without altering core state.
No external state management dependencies are required. The routing module reads existing state fields such as verdict, recovery_mode_consecutive_count, iteration, and max_iterations to compute routes and emit log events.
cursorLoop user task prompts back into the agent for iterative refinement.
tankygranny05Distinguish ephemeral warmup SIDs from persistent session SIDs in sse_lines.jsonl.
gweslabEnforce tool-based numeric computation for large numbers and complex arithmetic.
finsilabsImplement stackable price rules with priority ordering and customer segmentation.
