fix: --swarm flag silently ignored in /research and /research-deep#33
Merged
fix: --swarm flag silently ignored in /research and /research-deep#33
Conversation
The LLM was skipping advisory "Mode Selection" and executing Standard Workflow regardless of --swarm flag. Replaced with imperative "CRITICAL: Route Selection" gates and reordered Swarm Workflow before Standard Workflow to prevent top-to-bottom execution bias.
iamladi
added a commit
that referenced
this pull request
Apr 20, 2026
…main-model/CONTEXT-FORMAT.md's Rules section, converting the authored-output template's meta-instructions from directives into decidable guidance while preserving the bullet contract shape.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
/research: Replaced advisory "Argument Parsing" + "Mode Selection" with imperative "CRITICAL: Route Selection" gate; reordered Swarm Workflow before Standard Workflow/research-deep: Strengthened "CRITICAL: Parse Flags and Route" with imperative skip directives; reordered Swarm Discovery before Standard DiscoveryRoot Cause
The LLM read the command prompt top-to-bottom and started executing the first actionable section (Standard Workflow) before processing the routing gate. The weak advisory language ("If user requested swarm mode, execute Swarm Workflow below") wasn't treated as a hard stop.
Test plan
/research --swarm topictriggers agent team creation and spawns 3 teammates (Locator, Analyzer, Pattern Finder)/research topicruns standard subagent workflow with no team behavior/research-deep --swarm topictriggers swarm discovery with agent team/research-deep topicruns standard discovery