@skyf0xx/hedgehog 4.0.1 → 4.0.3

package/README.md

@@ -60,6 +60,14 @@ npx @skyf0xx/hedgehog graph # show graph
 
 ![The Hedgehog build graph](https://raw.githubusercontent.com/skyf0xx/hedgehog/master/docs/images/graph.png)
 
+## Parallel by Default
+
+Every dependency is explicit, so Hedgehog knows which tasks can run in parallel.
+
+![Comparison](https://raw.githubusercontent.com/skyf0xx/hedgehog/master/docs/images/comparison.png)
+
+Agents fan out to give you great outcomes at **faster speeds**.
+
 ## What Hedgehog builds
 
 ### Full-stack applications

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@skyf0xx/hedgehog",
-  "version": "4.0.1",
+  "version": "4.0.3",
   "description": "Install the Hedgehog build discipline (agents + skills) into a repo, for Claude Code, Cursor, or Gemini CLI.",
   "type": "module",
   "repository": {

package/src/agents/backend-eng.md

@@ -86,6 +86,10 @@ needed.
 
 ## Constraints
 
+- Default to no comments. Add one only when the WHY is non-obvious — a
+  hidden constraint, a workaround for a specific bug, an invariant the
+  code alone can't convey. Never comment WHAT the code does; a
+  well-named schema field, function, or variable already says that.
 - Never self-certify a task as done or run `git commit` for its changes —
   see Workflow step 3.
 - Never import another module's repository, service, or schema directly

package/src/agents/front-end-eng.md

@@ -89,6 +89,10 @@ don't reach for a second one.
 
 ## Constraints
 
+- Default to no comments. Add one only when the WHY is non-obvious — a
+  hidden constraint, a workaround for a specific bug, an invariant the
+  code alone can't convey. Never comment WHAT the code does; a
+  well-named component, hook, or variable already says that.
 - Never self-certify a task as done or run `git commit` for its changes —
   see Workflow step 3.
 - Never add a data-fetching call that bypasses the hook/contract layer —

package/src/agents/layer-eng.md

@@ -68,6 +68,10 @@ parsing and typing, and the layer after it consumes the result.
 
 ## Constraints
 
+- Default to no comments. Add one only when the WHY is non-obvious — a
+  hidden constraint, a workaround for a specific bug, an invariant the
+  code alone can't convey. Never comment WHAT the code does; a
+  well-named symbol already says that.
 - Never self-certify a task as done or run `git commit` for its changes —
   see Workflow step 4.
 - Never write outside the packet's ALLOWED SCOPE. Scope is what stops

package/src/agents/planner.md

@@ -263,8 +263,10 @@ accounts get added where there were none).
    own archival planning output (`.hedgehog/BMAD/` or `.hedgehog/chain/`,
    first run only), the authored core's `.hedgehog/core.yaml` and
    `.hedgehog/core-design.md` if step 4 ran, and root `CLAUDE.md`'s filled
-   placeholders (first run only). This is planning intake's own unit of
-   work, landed before `bootstrap` touches anything.
+   placeholders (first run only). Write these with the `no-history-in-output`
+   skill: current state only, no narration of the intake conversation.
+   This is planning intake's own unit of work, landed before `bootstrap`
+   touches anything.
 9. **First run only — hand off to the `bootstrap` agent** once the
    commit lands. It scaffolds the chosen core's workspace (and, for
    full-stack-app, whichever add-ons are on) before any build step

package/src/skills/no-history-in-output/SKILL.md

@@ -0,0 +1,38 @@
+---
+name: no-history-in-output
+description: Apply whenever writing or updating a project-facing document — root CLAUDE.md, `.hedgehog/core-design.md`, specs, READMEs, or similar. Ensures the file reads as a clean, current, as-is snapshot rather than a log of edits or decisions. Applies on first generation and on every later revision, including "update," "rewrite," or "clean up" requests.
+---
+
+# No History in Output
+
+Project-facing documents describe the current state of the thing, not the
+process that produced it.
+
+## Rules
+
+- Never include labels like "(unchanged)", "(updated)", "(new)",
+  "(revised)", or similar edit-tracking annotations in headings or body
+  text.
+- Never narrate the discussion that led to a decision (e.g. "we
+  considered X but decided against it," "originally this was Y, now it's
+  Z," "per your feedback...").
+- Never include changelogs, version history, "what changed" sections, or
+  meta-commentary about the conversation, unless the user explicitly
+  asks for a changelog as a deliverable in its own right.
+- Write every section as a plain statement of fact about the current
+  design/plan/content, as if writing it fresh with full knowledge of the
+  final decision — not as a diff against a prior version.
+- If a section documents a resolved tradeoff or an intentional design
+  choice, state the choice and its rationale plainly (e.g. "X is an
+  accepted tradeoff because..."), without referencing that alternatives
+  were debated or rejected earlier in conversation.
+- An "update this doc" request re-derives a clean as-is document — it
+  does not layer edit notes onto the previous version.
+
+## When NOT to apply
+
+- The user explicitly asks for a changelog, revision history, or "show
+  me what changed."
+- The document's entire purpose is to record a process over time — a
+  friction log entry (`hedgehog friction add`), a commit message, or
+  meeting notes. Those describe what happened and when on purpose.

package/src/templates/graph.html

@@ -728,7 +728,6 @@
     const [selectedId, setSelectedId] = useState(null);
     const [chainRootId, setChainRootId] = useState(null);
     const [collapsedModules, setCollapsedModules] = useState(() => new Set());
-    const [autoCollapseApplied, setAutoCollapseApplied] = useState(false);
     // Tracks the last-rendered graph's JSON so a poll that returns
     // identical data (the common case — nothing changed since last
     // tick) skips relayout entirely instead of jittering node positions
@@ -765,10 +764,12 @@
     const isLargeGraph = !!raw && raw.nodes.length > LARGE_GRAPH_THRESHOLD;
 
     // The modules worth a first look — anything with a blocked task
-    // (needs a fix) or a ready task (actionable). Drives both which lanes
-    // start collapsed and, on first paint, which lanes fitView zooms to
-    // (see attentionModulesRef below) — a large graph should open on what
-    // needs attention, not zoomed out far enough to fit every idle lane.
+    // (needs a fix) or a ready task (actionable). Falls back to the first
+    // module alphabetically when nothing qualifies (a fresh build, every
+    // task still `planned`), so there's always a target. Drives which lane
+    // the initial viewport jump (below) lands on for a large graph — it
+    // should open on what needs attention, not zoomed out far enough to
+    // fit every idle lane.
     const attentionModules = useMemo(() => {
       if (!raw) return new Set();
       const byModule = new Map();
@@ -780,22 +781,14 @@
       for (const [moduleName, statuses] of byModule) {
         if (statuses.some((s) => s === 'blocked' || s === 'ready')) next.add(moduleName);
       }
+      if (next.size === 0 && byModule.size > 0) {
+        next.add([...byModule.keys()].sort((a, b) => a.localeCompare(b))[0]);
+      }
       return next;
     }, [raw]);
 
-    // Auto-collapse every lane except the ones in attentionModules, once,
-    // the first time a graph crosses the large-graph threshold. A user's
-    // own later clicks always win after that: this never re-collapses a
-    // lane the user chose to open.
-    useEffect(() => {
-      if (!raw || !isLargeGraph || autoCollapseApplied) return;
-      const next = new Set();
-      for (const n of raw.nodes) {
-        if (!attentionModules.has(n.module)) next.add(n.module);
-      }
-      setCollapsedModules(next);
-      setAutoCollapseApplied(true);
-    }, [raw, isLargeGraph, autoCollapseApplied, attentionModules]);
+    // Lanes always start open — collapsing is a choice the user makes by
+    // clicking a lane header (toggleLane), never applied on their behalf.
 
     const edgesBySource = useMemo(() => {
       const map = new Map();
@@ -897,19 +890,15 @@
     // corner of the first lane worth attention (or the first lane at all,
     // if none need attention) at a fixed, legible zoom — rather than
     // fitting/centering on a bounding box, which (see above) can leave
-    // the viewport scrolled past the top of the content it fit to. Gated
-    // on autoCollapseApplied (not just isLargeGraph) so this runs once
-    // `lanes` reflects the post-collapse layout, not the pre-collapse one
-    // — setting the viewport before collapse applies would target
-    // coordinates that shift out from under it a moment later.
+    // the viewport scrolled past the top of the content it fit to.
     useEffect(() => {
-      if (!isLargeGraph || !autoCollapseApplied || initialViewSetRef.current) return;
+      if (!isLargeGraph || initialViewSetRef.current) return;
       const instance = flowInstanceRef.current;
       if (!instance || lanes.length === 0) return;
       const targetLane = lanes.find((l) => attentionModules.has(l.module)) ?? lanes[0];
       instance.setViewport({ x: -targetLane.x + 40, y: -targetLane.y + 40, zoom: 1 });
       initialViewSetRef.current = true;
-    }, [isLargeGraph, autoCollapseApplied, lanes, attentionModules]);
+    }, [isLargeGraph, lanes, attentionModules]);
 
     if (!raw) {
       return h('div', { id: 'loading' }, 'Loading build graph…');