You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This PR does not adopt EvoDev wholesale. It applies the same planning insight to Humanize's gen-plan: every implementation task should know where it sits in the global capability/feature structure. The new map gives tasks a shared frame of reference before they are broken down into executable coding / analyze items.
After adding the map to plan generation, this also wires that context into RLCR execution so Claude coding prompts, Codex reviews, round contracts, and the Goal Tracker can keep work anchored to the relevant capability node instead of losing the global position during later rounds.
Summary
add a Feature Map / Capability Map section to the gen-plan output template
require gen-plan to preserve capability-level business/design/implementation context before Task Breakdown
document refine-plan preservation behavior for the optional capability map section
add RLCR Capability Anchor guidance to round-0 setup, follow-up prompts, drift recovery, and review phase prompts
add a Capability column to Goal Tracker Active Tasks generated by RLCR setup
add Codex Capability Map Alignment Check sections to regular and full alignment review templates
update the Claude-facing Humanize command/skill docs so the plugin exposes the new capability-anchor behavior
document capability maps and RLCR capability anchors in README and the Claude usage guide
keep Claude plugin metadata and version unchanged at 1.17.0
add structure tests for the new planning and RLCR execution contracts
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
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.
Background
This change is inspired by the paper EvoDev: An Iterative Feature-Driven Framework for End-to-End Software Development with LLM-based Agents. EvoDev uses a Feature Map to model feature dependencies and propagate business, design, and implementation context across iterations.
This PR does not adopt EvoDev wholesale. It applies the same planning insight to Humanize's
gen-plan: every implementation task should know where it sits in the global capability/feature structure. The new map gives tasks a shared frame of reference before they are broken down into executablecoding/analyzeitems.After adding the map to plan generation, this also wires that context into RLCR execution so Claude coding prompts, Codex reviews, round contracts, and the Goal Tracker can keep work anchored to the relevant capability node instead of losing the global position during later rounds.
Summary
Feature Map / Capability Mapsection to the gen-plan output templateTask BreakdownCapability Anchorguidance to round-0 setup, follow-up prompts, drift recovery, and review phase promptsCapabilitycolumn to Goal Tracker Active Tasks generated by RLCR setupCapability Map Alignment Checksections to regular and full alignment review templates1.17.0Validation
jq empty .claude-plugin/plugin.json .claude-plugin/marketplace.jsonbash tests/test-task-tag-routing.shbash tests/test-template-references.shbash tests/test-templates-comprehensive.shbash tests/robustness/test-setup-scripts-robustness.shbash tests/test-plan-file-hooks.shbash tests/test-gen-plan.shbash tests/test-refine-plan.shgit diff --check