- Rewrite "How It Works" with 6 clear steps: init → discuss → plan → execute → verify → complete
- Emphasize discuss-phase as where users shape implementation (feeds research + planning)
- Emphasize verify-work as where users confirm features actually work (auto-diagnoses failures)
- Add brownfield callout at top of How It Works instead of separate section
- Move discuss-phase, verify-work, new-milestone into Core Workflow commands
- Replace "Subagent Execution" with "Multi-Agent Orchestration" explaining thin orchestrator pattern
- Highlight 30-40% context usage even after thousands of lines of code
Phase directories are now created on-demand when discuss-phase or
plan-phase first touches a phase, rather than upfront by roadmapper.
Prevents duplicate folders when roadmapper names phases differently
than what discuss/plan-phase derives from the roadmap.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Wipe commands/gsd/ and get-shit-done/ before copying
- Delete only gsd-*.md agents (preserves user agents)
- Add confirmation step to /gsd:update with changelog preview
- Warn users about clean install behavior before updating
Add structured next-step routing consistent with other GSD commands.
Four routes: pass+continue, pass+milestone, issues+ready, issues+blocked.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Commands with <offer_next> sections were outputting backticks literally
instead of rendering them as inline code. Removed code block wrappers
and added explicit 'Output this markdown directly' instruction.
Affected: plan-phase, execute-phase, audit-milestone
Updated routing in:
- new-project.md
- create-roadmap.md (command and workflow)
- progress.md (routes B and C)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Convert statusline.sh and gsd-check-update.sh to Node.js
- Remove gsd-notify.sh (blocking popups, not true toasts)
- Remove --force-notify and --no-notify flags from installer
- Use `node` command for cross-platform hook execution
Fixes#114
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove useless "None — you decide" skip option
- Generate phase-specific gray areas based on domain analysis
(UI features, CLI tools, APIs, organization tasks, etc.)
- Increase probing depth: 4 questions per area before check
- Make context.md categories flexible (emerge from discussion)
- Add CLI and organization examples to context template
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Commands accept both padded (08) and unpadded (8) phase numbers.
Normalization happens in step 1/2 before any directory lookups.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Support both zero-padded (05-*) and unpadded (5-*) phase folders
- Delete orphaned research-subagent-prompt.md template
- Update resume-project.md to use Task tool resume parameter
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
/gsd:execute-phase handles single and multiple plans.
All references updated to use execute-phase.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Commands load agent expertise directly via Task tool spawning.
Thin orchestrator pattern — agents have methodology baked in.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Orchestrator sometimes makes small fixes between executor completions
(missing imports, wiring fixes). These were left uncommitted. Now
checks for dirty state after all waves complete and commits if needed.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Revision mode was missing git commit step - plans updated based on
checker issues were written to disk but never committed.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
STATE.md reflects project state, not session state. It persists across
sessions and shows misleading info like "Phase complete" from previous
work. Now only uses session-scoped todo list.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Researcher was only checking if CONTEXT.md exists (ls), not
actually reading it (cat). Added mandatory bash command to
read phase context file in step 1.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The hook was grepping for `current_phase:` and `status:` but STATE.md
uses `Phase:` and `Status:`. Now shows actual phase info instead of
always falling back to "Ready for input".
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Planner was only describing that it should use these files,
not explicitly reading them. Added mandatory bash commands
to cat phase-specific context files in gather_phase_context step.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Researcher now reads and respects user decisions from discuss-phase:
- Locked decisions constrain research scope
- Claude's discretion areas get options explored
- Deferred ideas ignored
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Research is now integrated into /gsd:plan-phase by default.
Remove standalone research command from "Also available" sections.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Alerts user when Claude stops (task complete, needs input, etc.)
with context about what completed. Works on Mac (osascript alert),
Linux (notify-send/zenity), and Windows (PowerShell MessageBox).
Follows same optional install pattern as statusline:
- Prompts in interactive mode if existing hook detected
- --force-notify to replace existing
- --no-notify to skip installation
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
