docs: Simon's critical architectural feedback#67
Merged
neoneye merged 5 commits intoPlanExeOrg:mainfrom Feb 25, 2026
Merged
Conversation
…ation - Gap 2: Remove false assumption about MCP task_retry; it's internal to llm_executor.py, not suitable for ReviewPlan bridge - Adaptation 2: Mark ReviewPlan → task_retry bridge as INVALID (recursive calls, DB corruption risk) - Adaptation 3: Acknowledge PlanExe already uses systemic structured output (JSON + Markdown dual output, premise_attack.py, identify_potential_levers.py, premortem.py) - Priority order: Focus on Fermi Sanity Check (Adaptation 5) first; remove Adaptations 2 and 3 from scope Changes per Simon Strandgaard feedback, 2026-02-25
… output, quantitative grounding) Documents Simon Strandgaard's non-negotiable constraints for all future PlanExe work: 1. Task retry logic is LOCAL to llm_executor.py, NOT exposed via MCP - Do not use MCP task_retry from run_plan_pipeline.py - Would cause recursive calls and database corruption - Retry coordination must use Luigi's existing task dependencies 2. Structured output validation is SYSTEMIC across PlanExe - Already implemented in premise_attack.py, identify_potential_levers.py, premortem.py, and many files - Pattern: Save both JSON (full context) and Markdown (pretty-printed) - Do not add task-level Pydantic validation as new pattern—follow existing systemic approach 3. Quantitative grounding is the PRIMARY quality signal - All estimates must have lower bound, upper bound, and evidence - Flag any estimate where upper/lower > 100x as ungrounded - Ungrounded numbers are the root cause of 2-order-of-magnitude failures This file must be read before any work on retry logic, output validation, or task orchestration. Applies to all future development. Created per Mark Barney request to capture Simon's feedback for persistent reference.
Simon requested keeping docs/proposals/69-arcgentica-agent-patterns.md (the main proposal) and deleting docs/SIMON_FEEDBACK_ARCHITECTURE.md (the separate architecture doc). The feedback is captured in the proposal updates; separate doc was redundant.
This reverts commit 5767d67.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Documents three non-negotiable architectural constraints for all future PlanExe development:
See SIMON_FEEDBACK_ARCHITECTURE.md for full constraints and implications.