fix: move feature folders under workspace/development/features/

Davidson moved workspace/features/ to workspace/development/features/
(keeping engineering artifacts grouped under development/) and asked
to update the prompt convention in every agent/rule that referenced
the old path.

Updated references:
- .claude/rules/dev-phases.md (canonical path definition)
- .claude/agents/compass-planner.md (PRD + plan location)
- .claude/agents/helm-conductor.md (glob target for sequencing)
- .claude/commands/helm-conductor.md (same)
- .claude/agents/mirror-retro.md (retro save location)
- docs/agents/engineering-layer.md (public docs)

Also bundles the terminal auto-reply fix — registers CSI handlers for
final 'c' (DA1/DA2), 'n' (DSR status + cursor position), and 't'
(window ops) on the xterm.js parser so those queries are silently
absorbed before xterm.js can generate a reply. Previous attempts
(cols/rows on spawn, regex filter) left "0?1;2c" / "000000" junk in
the prompt because some replies slipped through as fragmented bytes
or via the CSI-t window-op path. The parser-level block is the real
fix; the regex filter stays as a second line of defense with 't'/'c'
now included in the charset.

CLAUDE.md also got the same workspace path fix but is gitignored —
personal workspace file, stays local.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: DavidsonGomes

.claude/agents/compass-planner.md

@@ -46,7 +46,7 @@ You produce **two artifacts** in Phase 2 of the canonical workflow (see `.claude
 
 **Where to save:**
 
-- **Feature-scoped work** (non-trivial, named feature): save both to `workspace/features/{feature-slug}/`
+- **Feature-scoped work** (non-trivial, named feature): save both to `workspace/development/features/{feature-slug}/`
   - `[C]prd-{feature}.md`
   - `[C]plan-{feature}.md`
 - **Standalone/one-off work** (no feature name, small scope): save to `workspace/development/plans/[C]plan-{name}-{YYYY-MM-DD}.md` and skip the PRD
@@ -63,9 +63,9 @@ Use the template at `.claude/templates/dev-work-plan.md` for the plan. The PRD s
 |---|---|---|---|
 | Typo, rename, tiny bug | ❌ | ❌ (go direct to Bolt) | — |
 | Small bug fix, clear repro | ❌ | ✅ minimal | `workspace/development/plans/` |
-| Feature with clear acceptance criteria | ✅ (short) | ✅ | `workspace/features/{slug}/` |
-| New feature with ambiguity | ✅ (full) | ✅ | `workspace/features/{slug}/` |
-| High-stakes migration | ✅ (full) + RALPLAN-DR | ✅ | `workspace/features/{slug}/` |
+| Feature with clear acceptance criteria | ✅ (short) | ✅ | `workspace/development/features/{slug}/` |
+| New feature with ambiguity | ✅ (full) | ✅ | `workspace/development/features/{slug}/` |
+| High-stakes migration | ✅ (full) + RALPLAN-DR | ✅ | `workspace/development/features/{slug}/` |
 
 **Rule:** when in doubt, produce a short PRD. A 10-line PRD is infinitely better than a missing one.
 
@@ -115,13 +115,13 @@ Use the template at `.claude/templates/dev-work-plan.md` for the plan. The PRD s
 
 1. Always read your memory folder first: `.claude/agent-memory/compass-planner/`
 2. Read `.claude/rules/dev-phases.md` — you are the owner of Phase 2 (Planning)
-3. Check for a feature folder: is there a `workspace/features/{slug}/` for this work? If yes, read any prior artifacts (discovery, etc.) to inherit context
+3. Check for a feature folder: is there a `workspace/development/features/{slug}/` for this work? If yes, read any prior artifacts (discovery, etc.) to inherit context
 4. Classify intent: Trivial / Refactoring / Build from Scratch / Mid-sized
 5. For codebase facts, spawn `@scout-explorer` (in parallel with other research)
 6. Ask user only about: priorities, timelines, scope decisions, risk tolerance, preferences
 7. For non-trivial work, consult `@echo-analyst` first for gap analysis
 8. Produce the PRD first (if applicable), then the plan derived from it
-9. Save to the feature folder (`workspace/features/{slug}/`) or `workspace/development/plans/` per the table above
+9. Save to the feature folder (`workspace/development/features/{slug}/`) or `workspace/development/plans/` per the table above
 10. Display confirmation summary, wait for explicit "proceed"
 11. On approval, hand off to `@apex-architect` (Phase 3) for non-trivial features, or directly to `@bolt-executor` (Phase 4) for clear executable plans
 12. Update agent memory with patterns worth remembering

.claude/agents/helm-conductor.md

@@ -37,11 +37,11 @@ Read `dev-phases.md` at the start of every session. Your recommendations must al
 
 You don't have a dedicated working folder — you don't produce artifacts. You read from:
 
-- `workspace/features/` — all active feature folders
+- `workspace/development/features/` — all active feature folders
 - `workspace/development/plans/` — standalone plans not yet in feature folders
 - `workspace/development/stories/` — story files (if present)
 
-When you need to record a sequencing decision, append it to `workspace/features/{feature}/[C]helm-notes.md` in the relevant feature — short entries only.
+When you need to record a sequencing decision, append it to `workspace/development/features/{feature}/[C]helm-notes.md` in the relevant feature — short entries only.
 
 ## Identity
 
@@ -51,7 +51,7 @@ When you need to record a sequencing decision, append it to `workspace/features/
 
 ## How you operate
 
-1. **Read first, recommend second.** Before answering "what next", glob `workspace/features/*/` and read the most recent artifact in each (discovery, PRD, plan, verification, retro). You cannot sequence what you haven't seen.
+1. **Read first, recommend second.** Before answering "what next", glob `workspace/development/features/*/` and read the most recent artifact in each (discovery, PRD, plan, verification, retro). You cannot sequence what you haven't seen.
 
 2. **Respect the 6 phases.** Every task belongs to a phase (Discovery, Planning, Solutioning, Build, Verify, Retro). Name the phase when you recommend.
 
@@ -80,7 +80,7 @@ For the full rules, entry/exit criteria and skip conditions, read `.claude/rules
 
 ### "What should I work on next?"
 
-1. Read all `workspace/features/*/` folders. For each, determine the current phase (look at which artifacts exist).
+1. Read all `workspace/development/features/*/` folders. For each, determine the current phase (look at which artifacts exist).
 2. For each feature, identify the next action (next phase or a blocker).
 3. Rank by: blockers first (to unblock), then by dependency order, then by priority signal from memory.
 4. Recommend the top 1-3 with phase + owner + why.
@@ -93,7 +93,7 @@ For the full rules, entry/exit criteria and skip conditions, read `.claude/rules
 
 ### "Sprint planning for {period}"
 
-1. List candidate features (from `workspace/features/` or the user).
+1. List candidate features (from `workspace/development/features/` or the user).
 2. For each, identify which phases remain.
 3. Order by dependency.
 4. Propose a sequence: "Week 1: Feature A Phase 2-3, Feature B Phase 4. Week 2: Feature A Phase 4, Feature C Phase 1."

.claude/agents/mirror-retro.md

@@ -37,7 +37,7 @@ Beyond your own agent memory in `.claude/agent-memory/mirror-retro/`, you have *
 
 Your workspace folder: the **feature folder itself**. You write retros next to the other artifacts:
 
-- `workspace/features/{feature-slug}/[C]retro-{feature}-{YYYY-MM-DD}.md`
+- `workspace/development/features/{feature-slug}/[C]retro-{feature}-{YYYY-MM-DD}.md`
 
 For sprint-level retros spanning multiple features:
 

.claude/commands/helm-conductor.md

@@ -1,3 +1,3 @@
 Use the @helm-conductor agent to orchestrate the engineering cycle — sequence features, decide what to work on next, route tasks to the right specialist agent, or plan a sprint: $ARGUMENTS
 
-Helm is the conductor, not a worker. It reads the state of `workspace/features/`, understands the 6-phase workflow in `.claude/rules/dev-phases.md`, and recommends the next concrete action with owner agent, expected output, and blockers. If no arguments are provided, Helm should read the feature folders and recommend what to work on next.
+Helm is the conductor, not a worker. It reads the state of `workspace/development/features/`, understands the 6-phase workflow in `.claude/rules/dev-phases.md`, and recommends the next concrete action with owner agent, expected output, and blockers. If no arguments are provided, Helm should read the feature folders and recommend what to work on next.

.claude/rules/dev-phases.md

@@ -84,7 +84,7 @@ Call `@helm-conductor` when you have multiple active features or are unsure whic
 Every non-trivial piece of work lives in its own feature folder:
 
 ```
-workspace/features/{feature-slug}/
+workspace/development/features/{feature-slug}/
 ├── [C]discovery-{feature}.md       ← Phase 1 (Echo)
 ├── [C]prd-{feature}.md             ← Phase 2 (Compass)
 ├── [C]plan-{feature}.md            ← Phase 2 (Compass)
@@ -110,7 +110,7 @@ workspace/features/{feature-slug}/
 
 Before starting work, every engineering agent should check for prior artifacts in the active feature folder:
 
-1. **Is there a feature folder for this work?** Look for `workspace/features/{slug}/`
+1. **Is there a feature folder for this work?** Look for `workspace/development/features/{slug}/`
 2. **If yes**, read in order: discovery → PRD → plan → architecture → any prior reviews/verifications
 3. **Inherit** constraints, decisions, and open questions — don't re-litigate them
 4. **If unclear** which feature this is, ask the user or call `@helm-conductor`
@@ -129,7 +129,7 @@ When one agent hands off to another, the handoff includes:
 - **Expected output:** what the next agent should produce and where
 
 Example handoff:
-> "Compass → Bolt: plan saved to `workspace/features/dark-mode/[C]plan-dark-mode.md`. Architecture pending (Apex). Open question: token storage strategy (see plan §Open Questions). Expected: implementation against steps 1-5 + self-verification."
+> "Compass → Bolt: plan saved to `workspace/development/features/dark-mode/[C]plan-dark-mode.md`. Architecture pending (Apex). Open question: token storage strategy (see plan §Open Questions). Expected: implementation against steps 1-5 + self-verification."
 
 ---
 

dashboard/frontend/src/components/AgentTerminal.tsx

@@ -62,6 +62,28 @@ export default function AgentTerminal({ agent, workingDir, accentColor = '#00FFA
     termRef.current = term
     fitRef.current = fit
 
+    // Silence terminal query replies at the parser level — before
+    // xterm.js gets a chance to generate them. The pty already knows
+    // its own capabilities; forwarding emulator-side replies made
+    // claude see them as keyboard input and print bytes like "0?1;2c"
+    // or "000000" into the prompt on startup.
+    //
+    // Registering a handler that returns `true` marks the CSI as
+    // "handled" and prevents the default sendDeviceAttributesPrimary /
+    // sendDeviceAttributesSecondary / deviceStatus / reportWindow*
+    // paths from firing. No reply is emitted at all.
+    //
+    // - final 'c'            → DA1 (\x1b[c) and DA2 (\x1b[>c)
+    // - final 'n'            → DSR status (\x1b[5n) and cursor pos (\x1b[6n)
+    // - final 't'            → window manipulation reports (xterm
+    //                          CSI Ps ; Ps ; Ps t)
+    const noReply = () => true
+    term.parser.registerCsiHandler({ final: 'c' }, noReply)
+    term.parser.registerCsiHandler({ final: 'c', prefix: '>' }, noReply)
+    term.parser.registerCsiHandler({ final: 'n' }, noReply)
+    term.parser.registerCsiHandler({ final: 'n', prefix: '?' }, noReply)
+    term.parser.registerCsiHandler({ final: 't' }, noReply)
+
     const onResize = () => {
       try {
         fit.fit()
@@ -76,7 +98,15 @@ export default function AgentTerminal({ agent, workingDir, accentColor = '#00FFA
     }
     window.addEventListener('resize', onResize)
 
+    // Second line of defense: even though the parser-level handlers
+    // above should prevent every known query reply, drop any onData
+    // payload that still looks like a terminal auto-reply. Real user
+    // keyboard input (arrows \x1b[A-D, Home/End \x1b[H/F, function
+    // keys \x1b[<n>~, modified arrows \x1b[1;2A) don't match either
+    // alternative.
+    const AUTO_REPLY_RE = /^\x1b\[(\?|>)[0-9;]*[a-zA-Z]$|^\x1b\[[0-9;]*[nRct]$/
     term.onData((data) => {
+      if (AUTO_REPLY_RE.test(data)) return
       if (wsRef.current?.readyState === WebSocket.OPEN) {
         wsRef.current.send(JSON.stringify({ type: 'input', data }))
       }

docs/agents/engineering-layer.md

@@ -84,7 +84,7 @@ Discovery → Planning → Solutioning → Build → Verify → Retro
 
 **Cycle orchestration** — `@helm-conductor` sits above the phases and answers "what next?", "who does it?", and "what's blocked?". Call Helm when you have multiple active features or need sprint sequencing.
 
-**Feature folders** — non-trivial work lives in `workspace/features/{feature-slug}/` where all artifacts of one feature are grouped together. Standalone artifacts continue to live in `workspace/development/{type}/`.
+**Feature folders** — non-trivial work lives in `workspace/development/features/{feature-slug}/` where all artifacts of one feature are grouped together. Standalone artifacts continue to live in `workspace/development/{type}/`.
 
 ## The 25 `dev-*` Skills