feat: PROVING GROUND — dogfood validation milestone

CHANGELOG.md

@@ -5,6 +5,19 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased] — PROVING GROUND
+
+### Added
+
+- **`coverage` view** — Code-to-spec gap analysis: identifies `crate:`/`module:`/`pkg:` nodes lacking `implements` edges to `spec:`/`adr:` targets. Returns `meta.linked`, `meta.unlinked`, and `meta.coveragePct`
+- **Echo ecosystem seed fixture** — `test/fixtures/echo-seed.yaml` with 55 nodes and 70 edges for integration testing (5 milestones, 5 specs, 5 ADRs, 5 docs, 15 crates, 11 tasks, 9 issues)
+- **PROVING GROUND integration tests** — `test/proving-ground.test.js` validates 5 real project management questions against the Echo seed with deterministic ground truth
+- **Dogfood session transcript** — `docs/dogfood-session.md` documents CLI walkthrough of all 5 questions with answers and timing
+
+### Changed
+
+- **Test count** — 162 tests across 9 files (was 143 across 8)
+
 ## [2.0.0-alpha.1] - 2026-02-11
 
 ### Added

docs/dogfood-session.md

@@ -0,0 +1,147 @@
+# PROVING GROUND — Dogfood Session Transcript
+
+> **Date**: 2026-02-11
+> **Seed**: `test/fixtures/echo-seed.yaml` — Echo ecosystem (55 nodes, 70 edges)
+> **Runtime**: Node.js 20, vitest 3.2.4
+
+## Setup
+
+```bash
+$ git mind init
+Initialized empty git-mind graph
+
+$ git mind import test/fixtures/echo-seed.yaml
+Imported 55 nodes, 70 edges (144ms)
+```
+
+---
+
+## Q1: What blocks milestone M2?
+
+**View**: `milestone`
+
+```bash
+$ git mind view milestone --json | jq '.meta.milestoneStats["milestone:M2"]'
+{
+  "total": 2,
+  "done": 0,
+  "pct": 0,
+  "blockers": ["issue:E-003", "issue:E-004"]
+}
+```
+
+**Answer**: Two issues block M2 — `issue:E-003` (REST schema breaks on nested objects) blocks `task:E-007`, and `issue:E-004` (validation rejects valid payloads) blocks `task:E-008`. Both M2 tasks are incomplete (0%).
+
+**Timing**: <1ms
+
+---
+
+## Q2: Which ADRs lack implementation?
+
+**View**: `traceability`
+
+```bash
+$ git mind view traceability --json | jq '.meta.gaps | map(select(startswith("adr:")))'
+[
+  "adr:003-encryption-at-rest",
+  "adr:004-rest-vs-grpc"
+]
+```
+
+**Answer**: 2 of 5 ADRs have no `implements` edge pointing at them. ADR-003 (encryption at rest for PII) and ADR-004 (REST over gRPC) are decisions with no code backing. The other 3 ADRs are covered: `echo-core` implements ADR-001, `echo-db` implements ADR-002, `echo-api` implements ADR-005.
+
+**Timing**: <1ms
+
+---
+
+## Q3: Which crates are unlinked to specs?
+
+**View**: `coverage` (new in PROVING GROUND)
+
+```bash
+$ git mind view coverage --json | jq '.meta'
+{
+  "linked": [
+    "crate:echo-core",
+    "crate:echo-api",
+    "crate:echo-db",
+    "crate:echo-auth",
+    "crate:echo-queue",
+    "crate:echo-web"
+  ],
+  "unlinked": [
+    "crate:echo-crypto",
+    "crate:echo-cli",
+    "crate:echo-config",
+    "crate:echo-log",
+    "crate:echo-test-utils",
+    "crate:echo-bench",
+    "crate:echo-migrate",
+    "crate:echo-plugin",
+    "crate:echo-sdk"
+  ],
+  "coveragePct": 40
+}
+```
+
+**Answer**: 9 of 15 crates have no `implements` edge to any spec or ADR. Coverage is 40%. The unlinked crates are utility/tooling crates (crypto, cli, config, log, test-utils, bench, migrate, plugin, sdk) — they may not need formal specs, but the gap is now visible.
+
+**Timing**: <1ms
+
+---
+
+## Q4: What should a new engineer read first?
+
+**View**: `onboarding`
+
+```bash
+$ git mind view onboarding --json | jq '.meta.readingOrder | map(select(startswith("doc:")))'
+[
+  "doc:getting-started",
+  "doc:architecture-overview",
+  "doc:api-reference",
+  "doc:contributing",
+  "doc:deployment-guide"
+]
+```
+
+**Answer**: Start with `doc:getting-started` (the root — no dependencies). Then `doc:architecture-overview` (depends on getting-started). After that, `doc:api-reference` and `doc:contributing` (both depend on architecture-overview). Finally `doc:deployment-guide` (depends on getting-started, so it could be read earlier, but topological sort with alphabetical tie-breaking places it last).
+
+**Timing**: <1ms
+
+---
+
+## Q5: What's low-confidence?
+
+**View**: `suggestions` + `computeStatus()`
+
+```bash
+$ git mind view suggestions --json | jq '.edges[] | {from, to, label, confidence: .props.confidence}'
+{ "from": "issue:E-016", "to": "issue:E-017",          "label": "relates-to", "confidence": 0.2 }
+{ "from": "issue:E-018", "to": "crate:echo-crypto",    "label": "relates-to", "confidence": 0.3 }
+{ "from": "issue:E-019", "to": "spec:003-web-sockets",  "label": "relates-to", "confidence": 0.4 }
+{ "from": "issue:E-020", "to": "doc:contributing",       "label": "relates-to", "confidence": 0.3 }
+
+$ git mind status --json | jq '.health.lowConfidence'
+4
+```
+
+**Answer**: 4 edges have confidence below 0.5 (the low-confidence threshold). All are `relates-to` edges — likely AI-suggested links that haven't been reviewed. Confidence values range from 0.2 to 0.4.
+
+**Timing**: <1ms
+
+---
+
+## Summary
+
+| # | Question | View | Answer | Time |
+|---|----------|------|--------|------|
+| 1 | What blocks M2? | `milestone` | `issue:E-003`, `issue:E-004` | <1ms |
+| 2 | ADRs lacking impl? | `traceability` | `adr:003-encryption-at-rest`, `adr:004-rest-vs-grpc` | <1ms |
+| 3 | Unlinked crates? | `coverage` | 9 of 15 (40% coverage) | <1ms |
+| 4 | Read first? | `onboarding` | `doc:getting-started` → `doc:architecture-overview` | <1ms |
+| 5 | Low-confidence? | `suggestions` | 4 edges (0.2–0.4) | <1ms |
+
+**Total query time**: ~1ms for all 5 questions against a 55-node, 70-edge graph.
+
+All 5 questions answered correctly from the graph alone — no manual inspection, no external tools, just views.

src/views.js

@@ -401,5 +401,50 @@ defineView('onboarding', (nodes, edges) => {
   };
 });
 
+defineView('coverage', (nodes, edges) => {
+  // Code-to-spec gap analysis: which code nodes lack implements edges to specs?
+  const codePrefixes = new Set(['crate', 'module', 'pkg']);
+  const specPrefixes = new Set(['spec', 'adr']);
+
+  const codeNodes = nodes.filter(n => {
+    const p = extractPrefix(n);
+    return p && codePrefixes.has(p);
+  });
+  const specNodes = new Set(
+    nodes.filter(n => {
+      const p = extractPrefix(n);
+      return p && specPrefixes.has(p);
+    })
+  );
+
+  // Code nodes that have at least one implements edge to a spec/adr
+  const linkedSet = new Set(
+    edges
+      .filter(e => e.label === 'implements' && specNodes.has(e.to))
+      .map(e => e.from)
+  );
+
+  const linked = codeNodes.filter(n => linkedSet.has(n));
+  const unlinked = codeNodes.filter(n => !linkedSet.has(n));
+
+  // Self-contained subgraph: edges where both endpoints are in the result
+  const resultNodes = new Set([...codeNodes, ...specNodes]);
+  const resultEdges = edges.filter(
+    e => e.label === 'implements' && resultNodes.has(e.from) && resultNodes.has(e.to)
+  );
+
+  return {
+    nodes: [...resultNodes],
+    edges: resultEdges,
+    meta: {
+      linked,
+      unlinked,
+      coveragePct: codeNodes.length > 0
+        ? Math.round((linked.length / codeNodes.length) * 100)
+        : 100,
+    },
+  };
+});
+
 // Capture built-in names after all registrations
 builtInNames = new Set(registry.keys());