gsd-pi 2.82.0 → 3.0.0

package/src/resources/extensions/gsd/prompts/refine-slice.md

@@ -64,10 +64,10 @@ Then:
 2. {{skillActivation}} Record the installed skills you expect executors to use in each task plan's `skills_used` frontmatter.
 3. Define slice-level verification: the objective stopping condition. Plan real test files with real assertions; for simple slices, executable commands are fine.
 4. For non-trivial slices, plan observability / proof level / integration closure, threat surface, and requirement impact. Omit entirely for simple slices.
-5. Decompose the slice into tasks that fit one context window each. Every task must have Why / Files / Do / Verify / Done-when, plus a task plan with description, steps, must-haves, verification, inputs (backtick-wrapped paths), and expected output (backtick-wrapped paths).
+5. Decompose the slice into tasks that fit one context window each. Every task passed to `gsd_plan_slice` must use the exact keys `taskId`, `title`, `description`, `estimate`, `files`, `verify`, `inputs`, `expectedOutput`, and optional `observabilityImpact`. Put Why / Do / Done-when detail in `description`. `files`, `inputs`, and `expectedOutput` must be JSON arrays of strings, even for one path (for example, `"expectedOutput": ["src/index.ts"]`, never `"expectedOutput": "src/index.ts"`).
 6. **Persist planning state through `gsd_plan_slice`.** Call it with the full payload. The tool writes to the DB and renders `{{outputPath}}` and `{{slicePath}}/tasks/T##-PLAN.md` automatically. Do NOT rely on direct `PLAN.md` writes.
 7. **Self-audit the plan.** If every task were completed exactly as written, the slice goal/demo should be true. Every must-have maps to a task. Inputs and Expected Output are backtick-wrapped file paths.
-8. If refinement produced structural decisions that diverge from the sketch, append them to `.gsd/DECISIONS.md`.
+8. If refinement produced structural decisions that diverge from the sketch, call `gsd_decision_save` for each; the tool persists the decision and regenerates `.gsd/DECISIONS.md`.
 9. {{commitInstruction}}
 
 The slice directory and tasks/ subdirectory already exist. Do NOT mkdir.

package/src/resources/extensions/gsd/prompts/rewrite-docs.md

@@ -23,7 +23,7 @@ An override was issued by the user that changes a fundamental decision or approa
    - For DECISIONS.md: append a new decision entry documenting the override and why. Do NOT delete prior decisions — mark them as superseded with a note.
    - For slice plans (S##-PLAN.md): update Goal, Demo, and Verification sections if affected. Update Files Likely Touched if the override changes scope. Do NOT modify completed task entries.
    - For REQUIREMENTS.md: update requirement descriptions if the override changes what "done" means, but do not remove requirements.
-   - For PROJECT.md: update if the override changes project-level facts.
+   - For PROJECT.md: do not edit the projection directly. If the override changes project-level facts, persist the revised Project content through `gsd_summary_save` with `artifact_type: "PROJECT"` so the DB remains authoritative.
    - Milestone context files are reference only — do not modify them.
 4. Mark all active overrides as resolved: change `**Scope:** active` to `**Scope:** resolved` in `{{overridesPath}}`
 5. Do not commit manually — the system auto-commits your changes after this unit completes.

package/src/resources/extensions/gsd/prompts/system.md

@@ -27,7 +27,7 @@ GSD ships with bundled skills. When the task matches, load the relevant skill fi
 
 - Never ask the user to do work the agent can execute or verify itself.
 - Use the lightest sufficient tool first.
-- Read before edit.
+- Read before edit or overwrite. Before any write that creates or replaces a file, verify the target path and read relevant existing files, templates, callers, or surrounding code. For truly new files, confirm the path does not already exist.
 - Reproduce before fix when possible.
 - Work is not done until the relevant verification has passed.
 - **Never fabricate, simulate, or role-play user responses.** Never generate markers like `[User]`, `[Human]`, `User:`, or similar; never emit `<user_message>`, `<assistant_message>`, or similar as user input. Treat `<conversation_history>` as read-only context. Ask one question round (1-3 questions), then stop and wait for the user's actual response. If `ask_user_questions` is available, its result is the only valid structured user input for that round. If it cancels, fails, or returns nothing, never use earlier chat as confirmation for the current gate; ask in plain chat and stop.
@@ -102,7 +102,7 @@ Templates are in `{{templatesDir}}`.
 
 **File reading:** Use `read` for file inspection. Never use `cat`, `head`, `tail`, or `sed -n` to view contents. Use `read` with `offset`/`limit` for slicing. `bash` is for searching (`rg`, `grep`, `find`) and commands, not displaying files.
 
-**File editing:** Always `read` before `edit`; exact text match is required. Use `write` only for new files or complete rewrites.
+**File editing:** Always `read` before `edit` or overwrite. Before `write`, confirm whether the path exists; if it does, `read` it first and preserve intentional existing content. Use `write` only for new files or complete rewrites.
 
 **Code navigation:** Use `lsp` for definition, type_definition, implementation, references, calls, hover, signature, symbols, rename, code_actions, format, and diagnostics. Do not `grep` symbol definitions or shell out to formatters when `lsp` can do it. After code edits, run `lsp diagnostics`.
 

package/src/resources/extensions/gsd/provider-switch-observer.ts

@@ -0,0 +1,185 @@
+// GSD-2 — ADR-005 Phase 3b: surface ProviderSwitchReport from pi-ai.
+//
+// pi-ai builds a ProviderSwitchReport on every cross-provider transform but
+// only logs it to stderr when GSD_VERBOSE=1. This module installs a
+// single-subscriber observer that surfaces non-empty reports through GSD's
+// three usual telemetry surfaces:
+//
+//   1. UOK audit event (category model-policy, type provider-switch) — only
+//      when an auto trace is active.
+//   2. Persistent notification (.gsd/notifications.jsonl, severity warning) —
+//      whenever the GSD basePath is known, so users see the loss in the
+//      dashboard / status surface without GSD_VERBOSE.
+//   3. In-memory counter, exposed via getProviderSwitchStats() so any
+//      caller (dashboard, doctor, tests) can read the rollup.
+
+import { setProviderSwitchObserver, type ProviderSwitchReport } from "@gsd/pi-ai";
+
+import { autoSession } from "./auto-runtime-state.js";
+import { appendNotification } from "./notification-store.js";
+import { buildAuditEnvelope, emitUokAuditEvent } from "./uok/audit.js";
+
+/** Rollup of cross-provider context transformations, grouped by trace. */
+export interface ProviderSwitchStats {
+  /** Total non-empty reports observed since process start (or last reset). */
+  totalSwitches: number;
+  /** Sum of every counted transformation across all reports. */
+  totals: {
+    thinkingBlocksDropped: number;
+    thinkingBlocksDowngraded: number;
+    toolCallIdsRemapped: number;
+    syntheticToolResultsInserted: number;
+    thoughtSignaturesDropped: number;
+  };
+  /** Per-trace breakdown. Key "interactive" covers reports outside auto-mode. */
+  byTrace: Record<string, TraceSwitchStats>;
+  /** The most recent non-empty report, if any. */
+  lastReport: ProviderSwitchReport | null;
+  /** ISO timestamp of the most recent non-empty report, if any. */
+  lastAt: string | null;
+}
+
+export interface TraceSwitchStats {
+  switches: number;
+  lastReport: ProviderSwitchReport;
+  lastAt: string;
+}
+
+const INTERACTIVE_TRACE_KEY = "interactive";
+
+interface MutableTraceStats {
+  switches: number;
+  lastReport: ProviderSwitchReport;
+  lastAt: string;
+}
+
+let installed = false;
+let totalSwitches = 0;
+const totals = {
+  thinkingBlocksDropped: 0,
+  thinkingBlocksDowngraded: 0,
+  toolCallIdsRemapped: 0,
+  syntheticToolResultsInserted: 0,
+  thoughtSignaturesDropped: 0,
+};
+const byTrace = new Map<string, MutableTraceStats>();
+let lastReport: ProviderSwitchReport | null = null;
+let lastAt: string | null = null;
+
+/** Format a one-line summary suitable for a notification message. */
+function summarize(report: ProviderSwitchReport): string {
+  const parts: string[] = [];
+  if (report.thinkingBlocksDropped > 0) parts.push(`${report.thinkingBlocksDropped} thinking dropped`);
+  if (report.thinkingBlocksDowngraded > 0) parts.push(`${report.thinkingBlocksDowngraded} thinking downgraded`);
+  if (report.toolCallIdsRemapped > 0) parts.push(`${report.toolCallIdsRemapped} tool ids remapped`);
+  if (report.syntheticToolResultsInserted > 0) parts.push(`${report.syntheticToolResultsInserted} synthetic tool results`);
+  if (report.thoughtSignaturesDropped > 0) parts.push(`${report.thoughtSignaturesDropped} thought signatures dropped`);
+  return `Provider switch ${report.fromApi} → ${report.toApi}: ${parts.join(", ")}`;
+}
+
+function recordReport(report: ProviderSwitchReport): void {
+  const now = new Date().toISOString();
+  totalSwitches += 1;
+  totals.thinkingBlocksDropped += report.thinkingBlocksDropped;
+  totals.thinkingBlocksDowngraded += report.thinkingBlocksDowngraded;
+  totals.toolCallIdsRemapped += report.toolCallIdsRemapped;
+  totals.syntheticToolResultsInserted += report.syntheticToolResultsInserted;
+  totals.thoughtSignaturesDropped += report.thoughtSignaturesDropped;
+  lastReport = report;
+  lastAt = now;
+
+  const traceKey = autoSession.currentTraceId ?? INTERACTIVE_TRACE_KEY;
+  const existing = byTrace.get(traceKey);
+  if (existing) {
+    existing.switches += 1;
+    existing.lastReport = report;
+    existing.lastAt = now;
+  } else {
+    byTrace.set(traceKey, { switches: 1, lastReport: report, lastAt: now });
+  }
+}
+
+function emitAudit(report: ProviderSwitchReport): void {
+  const traceId = autoSession.currentTraceId;
+  const basePath = autoSession.basePath;
+  if (!traceId || !basePath) return;
+  try {
+    emitUokAuditEvent(
+      basePath,
+      buildAuditEnvelope({
+        traceId,
+        category: "model-policy",
+        type: "provider-switch",
+        payload: {
+          fromApi: report.fromApi,
+          toApi: report.toApi,
+          thinkingBlocksDropped: report.thinkingBlocksDropped,
+          thinkingBlocksDowngraded: report.thinkingBlocksDowngraded,
+          toolCallIdsRemapped: report.toolCallIdsRemapped,
+          syntheticToolResultsInserted: report.syntheticToolResultsInserted,
+          thoughtSignaturesDropped: report.thoughtSignaturesDropped,
+        },
+      }),
+    );
+  } catch {
+    // Audit emission is best-effort. Counter + notification still fire.
+  }
+}
+
+function emitNotification(report: ProviderSwitchReport): void {
+  try {
+    appendNotification(summarize(report), "warning", "workflow-logger");
+  } catch {
+    // Notification persistence is best-effort.
+  }
+}
+
+function handleReport(report: ProviderSwitchReport): void {
+  recordReport(report);
+  emitAudit(report);
+  emitNotification(report);
+}
+
+/**
+ * Install the pi-ai observer. Idempotent — calling more than once is a no-op
+ * after the first install.
+ */
+export function installProviderSwitchObserver(): void {
+  if (installed) return;
+  setProviderSwitchObserver(handleReport);
+  installed = true;
+}
+
+/** Uninstall the observer. Intended for tests. */
+export function uninstallProviderSwitchObserver(): void {
+  setProviderSwitchObserver(undefined);
+  installed = false;
+}
+
+/** Read-only snapshot of the in-memory rollup. */
+export function getProviderSwitchStats(): ProviderSwitchStats {
+  const trace: Record<string, TraceSwitchStats> = {};
+  for (const [key, value] of byTrace) {
+    trace[key] = { switches: value.switches, lastReport: { ...value.lastReport }, lastAt: value.lastAt };
+  }
+  return {
+    totalSwitches,
+    totals: { ...totals },
+    byTrace: trace,
+    lastReport: lastReport ? { ...lastReport } : null,
+    lastAt,
+  };
+}
+
+/** Reset the in-memory rollup. Intended for tests. */
+export function _resetProviderSwitchStats(): void {
+  totalSwitches = 0;
+  totals.thinkingBlocksDropped = 0;
+  totals.thinkingBlocksDowngraded = 0;
+  totals.toolCallIdsRemapped = 0;
+  totals.syntheticToolResultsInserted = 0;
+  totals.thoughtSignaturesDropped = 0;
+  byTrace.clear();
+  lastReport = null;
+  lastAt = null;
+}

package/src/resources/extensions/gsd/queue-reorder-ui.ts

@@ -43,6 +43,7 @@ export async function showQueueReorder(
     const items = [...pending];
     let cursor = 0;
     let grabbed = false;
+    let scrollOffset = 0;
     let cachedLines: string[] | undefined;
     let validation: DependencyValidation;
 
@@ -154,8 +155,10 @@ export async function showQueueReorder(
 
       const ui = makeUI(theme, width);
       const lines: string[] = [];
+      const queueRows: string[] = [];
       const push = (...rows: string[][]) => { for (const r of rows) lines.push(...r); };
       const add = (s: string) => truncateToWidth(s, width);
+      let cursorQueueRow = 0;
 
       const headerText = grabbed ? "  Queue Reorder — Moving Item" : "  Queue Reorder";
       push(ui.bar(), ui.blank(), ui.header(headerText), ui.blank());
@@ -188,11 +191,13 @@ export async function showQueueReorder(
         const label = item.title && item.title !== item.id ? `${item.id}  ${item.title}` : item.id;
 
         if (isCursor && grabbed) {
-          lines.push(add(`  ${theme.fg("warning", `▸▸ ${num}. ${label}`)}`));
+          cursorQueueRow = queueRows.length;
+          queueRows.push(add(`  ${theme.fg("warning", `▸▸ ${num}. ${label}`)}`));
         } else if (isCursor) {
-          lines.push(add(`  ${theme.fg("accent", `${GLYPH.cursor} ${num}. ${label}`)}`));
+          cursorQueueRow = queueRows.length;
+          queueRows.push(add(`  ${theme.fg("accent", `${GLYPH.cursor} ${num}. ${label}`)}`));
         } else {
-          lines.push(add(`    ${theme.fg("text", `${num}. ${label}`)}`));
+          queueRows.push(add(`    ${theme.fg("text", `${num}. ${label}`)}`));
         }
 
         // depends_on annotations
@@ -201,36 +206,37 @@ export async function showQueueReorder(
           if (completedIds.has(dep)) continue;
           const pairKey = `${item.id}:${dep}`;
           if (violatedPairs.has(pairKey)) {
-            lines.push(add(`       ${theme.fg("warning", `${GLYPH.statusWarning} depends_on: ${dep} — auto-removed on confirm`)}`));
+            queueRows.push(add(`       ${theme.fg("warning", `${GLYPH.statusWarning} depends_on: ${dep} — auto-removed on confirm`)}`));
           } else if (redundantPairs.has(pairKey)) {
-            lines.push(add(`       ${theme.fg("dim", `↳ depends_on: ${dep} (redundant)`)}`));
+            queueRows.push(add(`       ${theme.fg("dim", `↳ depends_on: ${dep} (redundant)`)}`));
           } else {
-            lines.push(add(`       ${theme.fg("dim", `↳ depends_on: ${dep}`)}`));
+            queueRows.push(add(`       ${theme.fg("dim", `↳ depends_on: ${dep}`)}`));
           }
         }
 
         // Missing deps
         for (const v of validation.violations.filter(v => v.milestone === item.id && v.type === 'missing_dep')) {
-          lines.push(add(`       ${theme.fg("error", `${GLYPH.statusWarning} depends_on: ${v.dependsOn} (does not exist)`)}`));
+          queueRows.push(add(`       ${theme.fg("error", `${GLYPH.statusWarning} depends_on: ${v.dependsOn} (does not exist)`)}`));
         }
       }
 
       // Removed deps feedback
+      const trailingLines: string[] = [];
       if (removedDeps.length > 0) {
-        push(ui.blank());
+        trailingLines.push(...ui.blank());
         for (const r of removedDeps) {
-          lines.push(add(`  ${theme.fg("success", `${GLYPH.statusDone} Removed: ${r.milestone} depends_on ${r.dep}`)}`));
+          trailingLines.push(add(`  ${theme.fg("success", `${GLYPH.statusDone} Removed: ${r.milestone} depends_on ${r.dep}`)}`));
         }
       }
 
       // Circular warning
       const circ = validation.violations.find(v => v.type === 'circular');
       if (circ) {
-        push(ui.blank());
-        lines.push(add(`  ${theme.fg("error", `${GLYPH.statusWarning} ${circ.message}`)}`));
+        trailingLines.push(...ui.blank());
+        trailingLines.push(add(`  ${theme.fg("error", `${GLYPH.statusWarning} ${circ.message}`)}`));
       }
 
-      push(ui.blank());
+      trailingLines.push(...ui.blank());
 
       // Hints — context-sensitive based on grab state
       const hints: string[] = [];
@@ -250,7 +256,19 @@ export async function showQueueReorder(
       }
       hints.push("esc");
 
-      push(ui.hints(hints), ui.bar());
+      trailingLines.push(...ui.hints(hints), ...ui.bar());
+
+      const maxOverlayRows = Math.max(10, process.stdout.rows ? Math.floor(process.stdout.rows * 0.8) : 24);
+      const availableQueueRows = Math.max(1, maxOverlayRows - lines.length - trailingLines.length);
+      const maxScroll = Math.max(0, queueRows.length - availableQueueRows);
+      if (cursorQueueRow < scrollOffset) {
+        scrollOffset = cursorQueueRow;
+      } else if (cursorQueueRow >= scrollOffset + availableQueueRows) {
+        scrollOffset = cursorQueueRow - availableQueueRows + 1;
+      }
+      scrollOffset = Math.min(Math.max(scrollOffset, 0), maxScroll);
+
+      lines.push(...queueRows.slice(scrollOffset, scrollOffset + availableQueueRows), ...trailingLines);
 
       cachedLines = lines;
       return lines;

package/src/resources/extensions/gsd/smart-entry-routing.ts

@@ -0,0 +1,77 @@
+// GSD-2 — Smart entry routing decisions.
+// Pure route selection for /gsd guided wizard choices.
+
+export type SmartEntryIsolationMode = "worktree" | "branch" | "none";
+
+export type ActiveTaskChoice =
+  | "execute"
+  | "auto"
+  | "status"
+  | "milestone_actions";
+
+export type ActiveTaskRoute =
+  | {
+      kind: "auto-bootstrap";
+      verboseMode: false;
+      options?: {
+        step?: true;
+        milestoneLock?: string;
+      };
+    }
+  | {
+      kind: "guided-dispatch";
+      unitType: "execute-task";
+    }
+  | {
+      kind: "status";
+    }
+  | {
+      kind: "milestone-actions";
+    };
+
+export function resolveGuidedExecuteLaunchMode(
+  isolationMode: SmartEntryIsolationMode,
+): "auto-step" | "guided-dispatch" {
+  return isolationMode === "worktree" ? "auto-step" : "guided-dispatch";
+}
+
+export function resolveActiveTaskChoiceRoute(input: {
+  choice: ActiveTaskChoice;
+  isolationMode: SmartEntryIsolationMode;
+  milestoneId: string;
+}): ActiveTaskRoute {
+  if (input.choice === "auto") {
+    return {
+      kind: "auto-bootstrap",
+      verboseMode: false,
+    };
+  }
+
+  if (input.choice === "execute") {
+    if (resolveGuidedExecuteLaunchMode(input.isolationMode) === "auto-step") {
+      return {
+        kind: "auto-bootstrap",
+        verboseMode: false,
+        options: {
+          step: true,
+          milestoneLock: input.milestoneId,
+        },
+      };
+    }
+
+    return {
+      kind: "guided-dispatch",
+      unitType: "execute-task",
+    };
+  }
+
+  if (input.choice === "status") {
+    return { kind: "status" };
+  }
+
+  if (input.choice === "milestone_actions") {
+    return { kind: "milestone-actions" };
+  }
+
+  throw new Error(`Invalid ActiveTaskChoice: ${input.choice}`);
+}

package/src/resources/extensions/gsd/state-reconciliation/drift/merge-state.ts

@@ -23,6 +23,7 @@ import {
 } from "../../native-git-bridge.js";
 import type { GSDState } from "../../types.js";
 import { logError, logWarning } from "../../workflow-logger.js";
+import { isGsdWorktreePath } from "../../worktree-root.js";
 import type { DriftContext, DriftHandler, DriftRecord } from "../types.js";
 
 export type MergeReconcileResult = "clean" | "reconciled" | "blocked";
@@ -46,7 +47,13 @@ function resolveGitDir(basePath: string): string {
       return isAbsolute(gitDir) ? gitDir : resolve(basePath, gitDir);
     }
   } catch (err) {
-    logWarning("recovery", `gitdir resolution failed: ${getErrorMessage(err)}`);
+    const message = getErrorMessage(err);
+    logWarning("recovery", `gitdir resolution failed: ${message}`);
+    if (isGsdWorktreePath(basePath)) {
+      throw new Error(
+        `Worktree integrity failure: ${basePath} is not a valid git worktree (git rev-parse failed: ${message.split("\n")[0]}). Repair or recreate the worktree before retrying.`,
+      );
+    }
   }
 
   return join(basePath, ".git");

package/src/resources/extensions/gsd/state-reconciliation/drift/project-md.ts

@@ -1,14 +1,13 @@
 // Project/App: GSD-2
 // File Purpose: ADR-017 unregistered-milestone drift handler. Detects
 // milestones whose on-disk directory has meaningful content (ROADMAP/
-// CONTEXT/SUMMARY) but no DB row, then runs the markdown importer to
-// reconcile. PROJECT.md is the human-facing index — the importer's source
-// of truth is the .gsd/milestones/ directory tree.
+// CONTEXT/SUMMARY) but no DB row, then fails closed with an explicit recovery
+// instruction. Markdown hierarchy import is reserved for operator-controlled
+// migration/recovery commands, not automatic runtime reconciliation.
 
 import { existsSync } from "node:fs";
 
 import { getMilestone, isDbAvailable } from "../../gsd-db.js";
-import { migrateHierarchyToDb } from "../../md-importer.js";
 import { findMilestoneIds } from "../../milestone-ids.js";
 import { resolveMilestoneFile } from "../../paths.js";
 import type { GSDState } from "../../types.js";
@@ -46,20 +45,18 @@ export function detectUnregisteredMilestoneDrift(
 }
 
 /**
- * Repair: invoke the markdown importer. migrateHierarchyToDb walks the same
- * findMilestoneIds list the detector uses and INSERTs OR IGNOREs every
- * missing milestone (and its slices/tasks) — idempotent under cap=2 retry.
- *
- * Note: even though we receive one record at a time, the importer is a
- * project-wide operation. Repeated invocation across multiple drift records
- * in the same pass is wasteful but safe; a future optimization could
- * coalesce by checking whether the importer has already run this pass.
+ * Repair intentionally fails closed. The project-root DB is authoritative at
+ * runtime; markdown-only milestones must be imported through an explicit
+ * migration/recovery command so operators opt into changing canonical state.
  */
 export function repairUnregisteredMilestone(
-  _record: UnregisteredMilestoneDrift,
-  ctx: DriftContext,
+  record: UnregisteredMilestoneDrift,
+  _ctx: DriftContext,
 ): void {
-  migrateHierarchyToDb(ctx.basePath);
+  throw new Error(
+    `Milestone ${record.milestoneId} exists only as markdown projection. ` +
+      "Runtime reconciliation will not import markdown into the authoritative DB; run explicit GSD recovery/migration if this markdown should repopulate the database.",
+  );
 }
 
 export const unregisteredMilestoneHandler: DriftHandler<UnregisteredMilestoneDrift> = {

package/src/resources/extensions/gsd/state-reconciliation/drift/roadmap.ts

@@ -1,8 +1,8 @@
 // Project/App: GSD-2
 // File Purpose: ADR-017 roadmap-divergence drift handler. Detects mismatches
-// between ROADMAP.md (parsed slice sequence + depends declarations) and the
-// DB slice rows for that milestone, then reconciles via the markdown
-// importer plus an explicit junction-table sync.
+// between ROADMAP.md (parsed slice sequence, depends declarations, and
+// checkboxes) and the DB slice rows for that milestone, then re-renders the
+// ROADMAP projection from the authoritative DB rows.
 
 import { existsSync, readFileSync } from "node:fs";
 
@@ -10,12 +10,12 @@ import {
   getMilestone,
   getMilestoneSlices,
   isDbAvailable,
-  syncSliceDependencies,
 } from "../../gsd-db.js";
-import { migrateHierarchyToDb } from "../../md-importer.js";
+import { renderRoadmapFromDb } from "../../markdown-renderer.js";
 import { findMilestoneIds } from "../../milestone-ids.js";
 import { parseRoadmap } from "../../parsers-legacy.js";
 import { resolveMilestoneFile } from "../../paths.js";
+import { isClosedStatus } from "../../status-guards.js";
 import type { GSDState } from "../../types.js";
 import type { DriftContext, DriftHandler, DriftRecord } from "../types.js";
 
@@ -46,14 +46,20 @@ function milestoneHasDivergence(
 
   const dbSlices = getMilestoneSlices(milestoneId);
   const dbSliceMap = new Map(dbSlices.map((s) => [s.id, s]));
+  const roadmapSliceIds = new Set<string>();
 
   for (let i = 0; i < roadmap.slices.length; i++) {
     const roadmapSlice = roadmap.slices[i]!;
+    roadmapSliceIds.add(roadmapSlice.id);
     const expectedSequence = i + 1;
     const dbSlice = dbSliceMap.get(roadmapSlice.id);
     if (!dbSlice) return true; // Roadmap has a slice the DB doesn't.
     if (dbSlice.sequence !== expectedSequence) return true;
     if (!arraysEqual(dbSlice.depends, roadmapSlice.depends)) return true;
+    if (isClosedStatus(dbSlice.status) !== roadmapSlice.done) return true;
+  }
+  for (const dbSlice of dbSlices) {
+    if (!roadmapSliceIds.has(dbSlice.id)) return true;
   }
   return false;
 }
@@ -77,29 +83,15 @@ export function detectRoadmapDivergenceDrift(
 }
 
 /**
- * Repair a milestone's roadmap divergence:
- *   1. migrateHierarchyToDb upserts slice rows (sequence + depends JSON
- *      update via ON CONFLICT DO UPDATE).
- *   2. syncSliceDependencies updates the junction table per slice — the
- *      importer only writes the JSON column, not the relational view.
+ * Repair a milestone's roadmap divergence by regenerating the projection from
+ * DB rows. ROADMAP.md is a projection; runtime reconciliation must not import
+ * slice presence, sequence, dependencies, or checkbox state from markdown.
  */
-export function repairRoadmapDivergence(
+export async function repairRoadmapDivergence(
   record: RoadmapDivergenceDrift,
   ctx: DriftContext,
-): void {
-  migrateHierarchyToDb(ctx.basePath);
-
-  const roadmapPath = resolveMilestoneFile(ctx.basePath, record.milestoneId, "ROADMAP");
-  if (!roadmapPath || !existsSync(roadmapPath)) return;
-
-  try {
-    const roadmap = parseRoadmap(readFileSync(roadmapPath, "utf-8"));
-    for (const slice of roadmap.slices) {
-      syncSliceDependencies(record.milestoneId, slice.id, slice.depends);
-    }
-  } catch {
-    /* parse failure: detector will fire again next pass */
-  }
+): Promise<void> {
+  await renderRoadmapFromDb(ctx.basePath, record.milestoneId);
 }
 
 export const roadmapDivergenceHandler: DriftHandler<RoadmapDivergenceDrift> = {

package/src/resources/extensions/gsd/state.ts

@@ -591,7 +591,7 @@ async function handleAllSlicesDone(
       recentDecisions: [],
       blockers: [
         `Milestone ${activeMilestone.id} validation verdict is needs-remediation but all slices are complete. ` +
-          `Add remediation slices via gsd_reassess_roadmap or override the verdict manually.`,
+          `Add remediation slices via gsd_reassess_roadmap, or run \`/gsd verdict pass --rationale "..."\` to override.`,
       ],
       nextAction: `Resolve ${activeMilestone.id} remediation before proceeding.`,
       registry, requirements,
@@ -1314,7 +1314,7 @@ export async function _deriveStateImpl(
         recentDecisions: [],
         blockers: [
           `Milestone ${activeMilestone.id} validation verdict is needs-remediation but all slices are complete. ` +
-            `Add remediation slices via gsd_reassess_roadmap or override the verdict manually.`,
+            `Add remediation slices via gsd_reassess_roadmap, or run \`/gsd verdict pass --rationale "..."\` to override.`,
         ],
         nextAction: `Resolve ${activeMilestone.id} remediation before proceeding.`,
         registry,
@@ -1441,7 +1441,7 @@ export async function _deriveStateImpl(
     const summaryPath = resolveTaskFile(basePath, activeMilestone.id, activeSlice.id, t.id, "SUMMARY");
     if (summaryPath && existsSync(summaryPath)) {
       t.done = true;
-      logWarning("reconcile", `task ${activeMilestone.id}/${activeSlice.id}/${t.id} reconciled via SUMMARY on disk (#2514)`, { mid: activeMilestone.id, sid: activeSlice.id, tid: t.id });
+      logWarning("reconcile", `task ${activeMilestone.id}/${activeSlice.id}/${t.id} reconciled via SUMMARY on disk`, { mid: activeMilestone.id, sid: activeSlice.id, tid: t.id });
     }
   }
 

package/src/resources/extensions/gsd/status-guards.ts

@@ -25,3 +25,16 @@ export function isDeferredStatus(status: string): boolean {
 export function isInactiveStatus(status: string): boolean {
   return isClosedStatus(status) || isDeferredStatus(status);
 }
+
+/** Returns true when a prior milestone should not block dispatch ordering. */
+export function isSkippedForDispatch(status: string): boolean {
+  return isClosedStatus(status) || status === "parked" || isDeferredStatus(status);
+}
+
+/**
+ * Returns true when a milestone is future/backlog work (not currently executing).
+ * Includes legacy/project-specific alias "planned" for compatibility.
+ */
+export function isFutureMilestoneStatus(status: string): boolean {
+  return status === "pending" || status === "queued" || status === "planned";
+}

package/src/resources/extensions/gsd/templates/knowledge.md

@@ -1,7 +1,7 @@
 # Project Knowledge
 
-Append-only register of project-specific rules, patterns, and lessons learned.
-Agents read this before every unit. Add entries when you discover something worth remembering.
+Project-specific rules plus projected patterns and lessons learned.
+Rules are maintained in this file. Patterns and lessons are persisted to the memories table and rendered here on the next session-start projection.
 
 ## Rules
 

package/src/resources/extensions/gsd/templates/plan.md

@@ -99,11 +99,11 @@
   - "Improve UI"
 
   Each task should usually include:
-  - Why: why this task exists / what part of the slice it closes
-  - Files: the main files likely touched
-  - Do: concrete implementation steps and important constraints
-  - Verify: the command, test, or runtime check that proves it worked
-  - Done when: a measurable acceptance condition
+  - description: why this task exists, concrete steps, and done-when acceptance
+  - files: JSON array of likely touched paths
+  - verify: the command, test, or runtime check that proves it worked
+  - inputs: JSON array of existing paths or prior task outputs this task consumes
+  - expectedOutput: JSON array of paths this task creates or overwrites
 
   Keep the checkbox line format exactly:
   - [ ] **T01: Title** `est:30m`
@@ -131,10 +131,14 @@
 
   Verify field rules:
   - MUST be a mechanically executable command: `npm test`, `grep -q "pattern" file`, `test -f path`
+  - MUST NOT use shell pipes, redirects, semicolons, backticks, command substitution, or output trimming
+  - MUST NOT use inline `node -e` assertions for verification; put assertions in a real test file and run it with `node --test` or a package test script
   - For content/document tasks: verify file existence, section count, YAML validity, or word count
     NOT exact phrasing, specific formulas, or "zero TBD" aspirational criteria
   - If no command can verify the output, write: "Manual review — file exists and is non-empty"
+  - BAD: `python3 -m pytest tests/ -q --tb=short 2>&1 | tail -5`
   - BAD: "Sections 3.1 and 3.2 exist with exact formulas. Zero TBD/TODO."
+  - GOOD: `python3 -m pytest tests/ -q --tb=short`
   - GOOD: `grep -c "^## " doc.md` returns >= 4 (4+ sections), `! grep -q "TBD\|TODO" doc.md`
 
   Integration closure rule: