diff --git a/.opencode/plugins/ecc-hooks.ts b/.opencode/plugins/ecc-hooks.ts
index e350adad6..86eaa7cdf 100644
--- a/.opencode/plugins/ecc-hooks.ts
+++ b/.opencode/plugins/ecc-hooks.ts
@@ -16,9 +16,26 @@
 import type { PluginInput } from "@opencode-ai/plugin"
 import * as fs from "fs"
 import * as path from "path"
+import { fileURLToPath, pathToFileURL } from "url"
 import changedFilesTool from "../tools/changed-files.ts"
 import dependencyAnalyzerTool from "../tools/dependency-analyzer.ts"
 
+/**
+ * Directory of this module.
+ *
+ * The plugin is ESM: tsc compiles it to .opencode/dist/plugins/*.js and OpenCode
+ * loads the .ts directly, so `__dirname` is not defined there. Reading it used to
+ * throw inside a try/catch and silently degrade the reported ECC version to the
+ * default. `import.meta.url` is the ESM source of truth; the `__dirname` check is
+ * kept first only so a CJS consumer of this module still resolves correctly.
+ */
+function resolvePluginDir(): string {
+  if (typeof __dirname === "string" && __dirname) return __dirname
+  return path.dirname(fileURLToPath(import.meta.url))
+}
+
+const PLUGIN_DIR = resolvePluginDir()
+
 /**
  * Type definitions for better type safety
  */
@@ -51,17 +68,84 @@ interface TodoEvent {
 }
 
 /**
- * Read ECC version from package.json
- * Falls back to a default if package.json cannot be read
+ * Artifacts that identify an ECC root, one per capability this plugin needs.
+ *
+ * These are required runtime files, never optional content: `skills/` belongs to
+ * the optional `workflow-quality` module, so probing for a skill rejected
+ * supported modular installs (e.g. `platform-configs` + `hooks-runtime` without
+ * `workflow-quality`) that still ship both the plugin and the tracker — silently
+ * disabling skill telemetry and the shell root export for those users.
+ *
+ * Each probe is the exact artifact the caller is about to use, so detection is
+ * layout detection rather than an authenticity control: a directory only
+ * qualifies because it contains what we are going to read.
  */
-function getECCVersion(): string {
-  try {
-    const packageJsonPath = path.resolve(__dirname, "../../package.json")
-    const packageJson = JSON.parse(fs.readFileSync(packageJsonPath, "utf-8"))
-    return packageJson.version || "2.0.0"
-  } catch {
-    return "2.0.0"
+const ECC_ROOT_PROBES = {
+  // scripts/lib/utils.js is the shared library every scripts/ module requires.
+  root: [path.join("scripts", "lib", "utils.js")],
+  // What the skill-run telemetry hook imports at runtime.
+  skillTracker: [path.join("scripts", "hooks", "skill-run-tracker.js")],
+  // What the resolver embedded in commands requires once they read
+  // CLAUDE_PLUGIN_ROOT, and the version lookup for shell.env.
+  resolver: [path.join("scripts", "lib", "resolve-ecc-root.js")],
+} as const
+
+/**
+ * Walk upwards from `startDir` looking for an ancestor that contains every
+ * relative path in `probes`.
+ *
+ * The plugin ships in three layouts — `<root>/plugins` (installed), `<root>/dist/plugins`
+ * (compiled) and `<repo>/.opencode/plugins` (repo checkout) — so the number of
+ * levels back to the root is not fixed. Probing beats a hardcoded `..`: a wrong
+ * root would make every command resolve against a directory that has no scripts.
+ *
+ * The walk is bounded to 5 levels so it cannot wander into unrelated ancestors.
+ *
+ * Returns null when no ancestor qualifies, so callers can leave the capability
+ * disabled instead of advertising a root that does not hold the artifact.
+ */
+function findEccRootDir(startDir: string, probes: readonly string[]): string | null {
+  let current = startDir
+  for (let depth = 0; depth < 5; depth += 1) {
+    if (probes.every((probe) => fs.existsSync(path.join(current, probe)))) {
+      return current
+    }
+    const parent = path.dirname(current)
+    if (parent === current) break
+    current = parent
   }
+  return null
+}
+
+/**
+ * Read the ECC version from package.json.
+ *
+ * Candidates are tried in order — the resolved ECC root first, then the two
+ * plausible parent depths of the plugin — because the layout differs between an
+ * installed root (`<root>/plugins`), the compiled bundle (`<root>/dist/plugins`)
+ * and a repo checkout (`<repo>/.opencode/plugins`). Falls back to a default if
+ * no package.json can be read.
+ */
+function getECCVersion(eccRootDir?: string | null): string {
+  const candidates = [
+    eccRootDir,
+    path.resolve(PLUGIN_DIR, ".."),
+    path.resolve(PLUGIN_DIR, "..", ".."),
+  ]
+
+  for (const candidate of candidates) {
+    if (!candidate) continue
+    try {
+      const packageJson = JSON.parse(fs.readFileSync(path.join(candidate, "package.json"), "utf-8"))
+      if (typeof packageJson.version === "string" && packageJson.version) {
+        return packageJson.version
+      }
+    } catch {
+      // Try the next candidate.
+    }
+  }
+
+  return "2.0.0"
 }
 
 type ECCHooksPluginFn = (input: PluginInput) => Promise<Record<string, unknown>>
@@ -135,6 +219,60 @@ export const ECCHooksPlugin: ECCHooksPluginFn = async ({
       .catch(() => {})
   }
 
+  // Claude Code routes post:skill:track through
+  // scripts/hooks/posttooluse-dispatcher.js, which OpenCode never invokes.
+  // Without a caller here, nothing ever writes skill-runs.jsonl and
+  // `skills-health.js --dashboard` has no data to report on any skill (#2463).
+  //
+  // The tracker module is reused rather than reimplemented so its identifier
+  // validation, length bounds and "never persist prompt text" guarantees stay
+  // the ones that were written and audited for the Claude Code path.
+  //
+  // Loaded lazily, for the same reason as the store above: this file is
+  // OpenCode's startup entry point, and a static import of a missing module
+  // crashes the plugin -- and with it the whole session -- before any hook can
+  // run (#2530). A runtime file URL also keeps tsc from resolving a relative
+  // specifier that exists in no compiled layout.
+  type SkillRunTrackerModule = { run?: (payload: unknown) => void }
+  let recordSkillRun: ((payload: unknown) => void) | undefined
+  // Detection is scoped to the tracker itself, so a modular install that omits
+  // optional content still gets telemetry.
+  const trackerRoot = findEccRootDir(PLUGIN_DIR, ECC_ROOT_PROBES.skillTracker)
+  // The root advertised to commands is a different question: those commands
+  // require the resolver, so a root without it would break them again.
+  const eccRoot =
+    findEccRootDir(PLUGIN_DIR, ECC_ROOT_PROBES.resolver) ?? findEccRootDir(PLUGIN_DIR, ECC_ROOT_PROBES.root)
+  try {
+    const trackerPath = trackerRoot
+      ? path.join(trackerRoot, "scripts", "hooks", "skill-run-tracker.js")
+      : null
+    if (!trackerPath || !fs.existsSync(trackerPath)) {
+      throw new Error("skill-run-tracker.js not found")
+    }
+    // CJS module loaded from ESM: run() is on the namespace in Node/Bun, but a
+    // transpiled default export must stay supported.
+    const loaded = (await import(pathToFileURL(trackerPath).href)) as SkillRunTrackerModule & {
+      default?: SkillRunTrackerModule
+    }
+    const entry = typeof loaded.run === "function" ? loaded : loaded.default
+    if (!entry || typeof entry.run !== "function") {
+      throw new Error("skill-run-tracker.js exposes no run()")
+    }
+    recordSkillRun = entry.run
+  } catch {
+    // Same deferred-log rationale as the store above: guarantee that a
+    // telemetry failure can never escape this catch and abort startup.
+    Promise.resolve()
+      .then(() =>
+        log(
+          "warn",
+          "[ECC] skill-run telemetry disabled: could not load scripts/hooks/skill-run-tracker.js. " +
+            "Run `ecc repair --target opencode` to restore the missing files. Other ECC hooks are unaffected."
+        )
+      )
+      .catch(() => {})
+  }
+
   const normalizeProfile = (value: string | undefined): HookProfile => {
     if (value === "minimal" || value === "strict") return value
     return "standard"
@@ -264,6 +402,25 @@ export const ECCHooksPlugin: ECCHooksPluginFn = async ({
       ) {
         log("info", "[ECC] PR created - check GitHub Actions status")
       }
+
+      // Skill telemetry (#2463). OpenCode's tool is named `skill`, so the
+      // harness-side name is matched here (case-insensitively) and the payload
+      // is handed over in the shape the tracker already understands.
+      if (
+        hookEnabled("post:skill:track", ["standard", "strict"]) &&
+        typeof input.tool === "string" &&
+        input.tool.trim().toLowerCase() === "skill"
+      ) {
+        try {
+          recordSkillRun?.({
+            tool_name: "Skill",
+            tool_input: input.args ?? {},
+            tool_response: output,
+          })
+        } catch {
+          // Telemetry is best-effort; never let it affect the tool result.
+        }
+      }
     },
 
     /**
@@ -483,7 +640,7 @@ export const ECCHooksPlugin: ECCHooksPluginFn = async ({
      */
     "shell.env": async (_input: { cwd: string }, output: { env: Record<string, string> }) => {
       const env: Record<string, string> = {
-        ECC_VERSION: getECCVersion(),
+        ECC_VERSION: getECCVersion(eccRoot),
         ECC_PLUGIN: "true",
         ECC_HOOK_PROFILE: currentProfile,
         ECC_DISABLED_HOOKS: process.env.ECC_DISABLED_HOOKS || "",
@@ -523,6 +680,25 @@ export const ECCHooksPlugin: ECCHooksPluginFn = async ({
         env.PRIMARY_LANGUAGE = detected[0]
       }
 
+      // resolve-ecc-root.js only probes ~/.claude and Claude Code's plugin
+      // cache, so under OpenCode it fell back to ~/.claude and every command
+      // carrying the inlined resolver failed to find its scripts. That resolver
+      // reads CLAUDE_PLUGIN_ROOT first, so exporting the root here fixes them
+      // through the mechanism ECC already provides -- rather than editing the
+      // command files that embed the locator.
+      //
+      // Only set when a root actually exists, and never replace a root someone
+      // else already chose: neither the process environment (set by the user)
+      // nor `output.env` (set by an earlier plugin). The final merge below
+      // overwrites output.env, so checking it only in `env` would clobber it.
+      const rootAlreadySet = Boolean(
+        (process.env.CLAUDE_PLUGIN_ROOT || "").trim() ||
+          (output.env?.CLAUDE_PLUGIN_ROOT || "").trim()
+      )
+      if (eccRoot && !rootAlreadySet) {
+        env.CLAUDE_PLUGIN_ROOT = eccRoot
+      }
+
       // OpenCode reads the supplied output object and ignores callback return values.
       output.env = { ...output.env, ...env }
     },
diff --git a/scripts/hooks/skill-run-tracker.js b/scripts/hooks/skill-run-tracker.js
index eef54b859..a306245d5 100644
--- a/scripts/hooks/skill-run-tracker.js
+++ b/scripts/hooks/skill-run-tracker.js
@@ -16,8 +16,9 @@
  * skill id is an identifier, not free text, so anything that does not look like
  * one is dropped rather than written through.
  *
- * Best-effort: never blocks tool execution. Runs under the PostToolUse
- * dispatcher, which owns stdin, pass-through, and exit codes.
+ * Best-effort: never blocks tool execution. Under Claude Code it runs via the
+ * PostToolUse dispatcher, which owns stdin, pass-through, and exit codes; under
+ * OpenCode the ecc-hooks plugin calls run() directly with an equivalent payload.
  *
  * Cross-platform (Windows, macOS, Linux); CommonJS.
  */
@@ -102,6 +103,28 @@ function deriveOutcome(payload) {
   return 'success';
 }
 
+const SKILL_TOOL_NAME = 'skill';
+
+/**
+ * Report whether a hook payload describes a skill tool invocation.
+ *
+ * Claude Code names the tool `Skill`; OpenCode and other harnesses name it
+ * `skill`. Comparing the raw string meant a case difference silently disabled
+ * the whole write side again — the hook ran, dropped every event, and the
+ * dashboard kept reporting unmeasured skills with no hint why.
+ *
+ * @param {{tool_name?: unknown} | null} payload
+ * @returns {boolean}
+ */
+function isSkillToolCall(payload) {
+  if (!payload || typeof payload !== 'object') {
+    return false;
+  }
+
+  const toolName = payload.tool_name;
+  return typeof toolName === 'string' && toolName.trim().toLowerCase() === SKILL_TOOL_NAME;
+}
+
 function buildRecord(payload) {
   const skillId = extractSkillId(payload.tool_input);
   if (!skillId) {
@@ -130,12 +153,22 @@ function buildRecord(payload) {
   };
 }
 
+/**
+ * Record a skill invocation from a hook payload.
+ *
+ * Accepts the raw JSON string Claude Code's dispatcher reads from stdin, or an
+ * already-parsed payload for in-process callers such as the OpenCode plugin.
+ * Non-skill tools and unparsable input are ignored.
+ *
+ * @param {string | object | null | undefined} rawInput
+ * @returns {void}
+ */
 function run(rawInput) {
   try {
     const payload = typeof rawInput === 'string'
       ? (rawInput.trim() ? JSON.parse(rawInput) : {})
       : rawInput;
-    if (payload && typeof payload === 'object' && payload.tool_name === 'Skill') {
+    if (isSkillToolCall(payload)) {
       const record = buildRecord(payload);
       if (record) {
         recordSkillExecution(record);
@@ -146,4 +179,4 @@ function run(rawInput) {
   }
 }
 
-module.exports = { buildRecord, deriveOutcome, extractSkillId, run };
+module.exports = { buildRecord, deriveOutcome, extractSkillId, isSkillToolCall, run };
diff --git a/scripts/lib/skill-evolution/dashboard.js b/scripts/lib/skill-evolution/dashboard.js
index 7049707bd..edd98ce74 100644
--- a/scripts/lib/skill-evolution/dashboard.js
+++ b/scripts/lib/skill-evolution/dashboard.js
@@ -328,6 +328,18 @@ function renderVersionTimelinePanel(skillsById, options = {}) {
   };
 }
 
+/**
+ * Render the full skill-health dashboard: header, coverage warning and panels.
+ *
+ * Coverage is stated explicitly because the 30-day window can empty out while
+ * runs are still retained, so "unmeasured" must not be reported as "never
+ * recorded" (#2463).
+ *
+ * @param {object} [options]
+ * @param {string} [options.now] ISO timestamp used as the evaluation instant
+ * @param {string} [options.panel] render only this panel
+ * @returns {{text: string, data: object}}
+ */
 function renderDashboard(options = {}) {
   const now = options.now || new Date().toISOString();
   const nowMs = Date.parse(now);
@@ -359,12 +371,31 @@ function renderDashboard(options = {}) {
   const header = [
     'ECC Skill Health Dashboard',
     `Generated: ${now}`,
-    `Skills: ${summary.total_skills} total, ${summary.healthy_skills} healthy, ${summary.declining_skills} declining`,
+    `Skills: ${summary.total_skills} total, ${summary.measured_skills} measured, ${summary.healthy_skills} healthy, ${summary.declining_skills} declining, ${summary.unmeasured_skills} unmeasured`,
     '',
   ];
 
   textParts.push(header.join('\n'));
 
+  if (summary.unmeasured_skills > 0) {
+    // The window can empty out while runs are still retained, so "unmeasured"
+    // must not be reported as "never recorded" (#2463). Three distinct states,
+    // because "no recent runs" alone does not say which one applies.
+    const recentRuns = health.filterRecordsWithinDays(records, nowMs, 30).length;
+    const coverageNote = records.length === 0
+      ? 'Nothing has ever been recorded, so the panels below are not a signal about skill health.'
+      : recentRuns === 0
+        ? `All ${records.length} retained run${records.length === 1 ? '' : 's'} are older than the 30-day window: coverage aged out rather than never existing. The panels below still show that history.`
+        : `Coverage is partial: ${recentRuns} recent run${recentRuns === 1 ? '' : 's'} inside the window, ${records.length - recentRuns} older one${records.length - recentRuns === 1 ? '' : 's'} retained outside it. Only the measured skills can be judged.`;
+
+    textParts.push([
+      `WARNING: ${summary.unmeasured_skills} of ${summary.total_skills} skills have no recorded run in the last 30 days.`,
+      'This dashboard can say nothing about them: no telemetry is an absence of evidence,',
+      'not a healthy result. Their success rate reads "n/a", which means unknown.',
+      coverageNote,
+    ].join('\n'));
+  }
+
   if (selectedPanel) {
     const result = panelRenderers[selectedPanel]();
     panels[selectedPanel] = result.data;
diff --git a/scripts/lib/skill-evolution/health.js b/scripts/lib/skill-evolution/health.js
index 309f16094..c57e23354 100644
--- a/scripts/lib/skill-evolution/health.js
+++ b/scripts/lib/skill-evolution/health.js
@@ -26,15 +26,36 @@ function formatRate(value) {
   return `${Math.round(value * 100)}%`;
 }
 
+/**
+ * Summarise a health report into counting buckets.
+ *
+ * A skill with no runs in the window is not healthy, it is unmeasured: the
+ * trend helpers return 'stable' when both rates are null, so deriving health
+ * from "not declining" turned a missing telemetry writer into a green
+ * dashboard across every skill (#2463). A declining skill always has runs, so
+ * the buckets partition the total.
+ *
+ * @param {{skills: Array<{declining?: boolean, run_count_30d?: number}>}} report
+ * @returns {{total_skills: number, measured_skills: number, healthy_skills: number, declining_skills: number, unmeasured_skills: number}}
+ */
 function summarizeHealthReport(report) {
   const totalSkills = report.skills.length;
+  // A skill with no runs in the window is not healthy, it is unmeasured.
+  // Folding it into "healthy" turned a missing telemetry writer into a green
+  // dashboard across every skill (#2463) — false confidence that never triggers
+  // an alert. A declining skill always has runs, so the buckets partition the
+  // total.
+  const unmeasuredSkills = report.skills.filter(skill => !skill.run_count_30d).length;
   const decliningSkills = report.skills.filter(skill => skill.declining).length;
-  const healthySkills = totalSkills - decliningSkills;
+  const measuredSkills = totalSkills - unmeasuredSkills;
+  const healthySkills = measuredSkills - decliningSkills;
 
   return {
     total_skills: totalSkills,
+    measured_skills: measuredSkills,
     healthy_skills: healthySkills,
     declining_skills: decliningSkills,
+    unmeasured_skills: unmeasuredSkills,
   };
 }
 
@@ -230,7 +251,7 @@ function formatHealthReport(report, options = {}) {
   const lines = [
     'ECC skill health',
     `Generated: ${report.generated_at}`,
-    `Skills: ${summary.total_skills} total, ${summary.healthy_skills} healthy, ${summary.declining_skills} declining`,
+    `Skills: ${summary.total_skills} total, ${summary.measured_skills} measured, ${summary.healthy_skills} healthy, ${summary.declining_skills} declining, ${summary.unmeasured_skills} unmeasured`,
     '',
     'skill            version   7d     30d    trend       pending   last run',
     '--------------------------------------------------------------------------',
diff --git a/tests/hooks/skill-run-tracker.test.js b/tests/hooks/skill-run-tracker.test.js
index d3e63e247..f836ab247 100644
--- a/tests/hooks/skill-run-tracker.test.js
+++ b/tests/hooks/skill-run-tracker.test.js
@@ -15,7 +15,7 @@ const fs = require('fs');
 const os = require('os');
 const path = require('path');
 
-const { buildRecord, deriveOutcome, extractSkillId, run } = require('../../scripts/hooks/skill-run-tracker');
+const { buildRecord, deriveOutcome, extractSkillId, isSkillToolCall, run } = require('../../scripts/hooks/skill-run-tracker');
 const { readHooksConfig } = require('../../scripts/lib/hooks-config');
 const {
   MAX_RUN_RECORDS,
@@ -152,6 +152,57 @@ test('run ignores non-Skill tools and malformed input without throwing', () => {
   assert.doesNotThrow(() => run(''));
 });
 
+// ── tool-name tolerance ───────────────────────────────────────────────────────
+// OpenCode names the tool `skill`. A case-sensitive matcher meant the hook ran,
+// dropped every event, and the dashboard kept reporting "unmeasured" with no
+// indication that a writer existed at all.
+
+test('isSkillToolCall accepts both harness spellings and rejects everything else', () => {
+  assert.strictEqual(isSkillToolCall({ tool_name: 'Skill' }), true);
+  assert.strictEqual(isSkillToolCall({ tool_name: 'skill' }), true);
+  assert.strictEqual(isSkillToolCall({ tool_name: '  SKILL  ' }), true);
+  assert.strictEqual(isSkillToolCall({ tool_name: 'Bash' }), false);
+  assert.strictEqual(isSkillToolCall({ tool_name: '' }), false);
+  assert.strictEqual(isSkillToolCall({ tool_name: 42 }), false);
+  assert.strictEqual(isSkillToolCall({}), false);
+  assert.strictEqual(isSkillToolCall(null), false);
+});
+
+test('run records a run for a lowercase skill tool name', () => {
+  withTempHome(homeDir => {
+    const previousHome = process.env.HOME;
+    const previousUserProfile = process.env.USERPROFILE;
+    process.env.HOME = homeDir;
+    process.env.USERPROFILE = homeDir;
+    try {
+      run(payload({ tool_name: 'skill' }));
+
+      const records = readSkillExecutionRecords({ homeDir });
+      assert.strictEqual(records.length, 1);
+      assert.strictEqual(records[0].skill_id, 'code-review');
+    } finally {
+      if (previousHome === undefined) delete process.env.HOME; else process.env.HOME = previousHome;
+      if (previousUserProfile === undefined) delete process.env.USERPROFILE; else process.env.USERPROFILE = previousUserProfile;
+    }
+  });
+});
+
+test('run still records nothing for a non-skill tool', () => {
+  withTempHome(homeDir => {
+    const previousHome = process.env.HOME;
+    const previousUserProfile = process.env.USERPROFILE;
+    process.env.HOME = homeDir;
+    process.env.USERPROFILE = homeDir;
+    try {
+      run(payload({ tool_name: 'bash' }));
+      assert.strictEqual(readSkillExecutionRecords({ homeDir }).length, 0);
+    } finally {
+      if (previousHome === undefined) delete process.env.HOME; else process.env.HOME = previousHome;
+      if (previousUserProfile === undefined) delete process.env.USERPROFILE; else process.env.USERPROFILE = previousUserProfile;
+    }
+  });
+});
+
 // ── JSONL sink bounds ────────────────────────────────────────────────────────
 
 test('recordSkillExecution writes the sink owner-only', function () {
diff --git a/tests/lib/skill-dashboard.test.js b/tests/lib/skill-dashboard.test.js
index f90052e85..d1f29d0df 100644
--- a/tests/lib/skill-dashboard.test.js
+++ b/tests/lib/skill-dashboard.test.js
@@ -358,6 +358,179 @@ function runTests() {
       }, /Unknown panel/);
     })) passed++; else failed++;
 
+    // A panel that reports every never-used skill as healthy is worse than an
+    // empty one: it looks like evidence and never raises an alert (#2463).
+    if (test('renderDashboard warns that unmeasured skills say nothing about health', () => {
+      const result = dashboard.renderDashboard({
+        skillsRoot,
+        learnedRoot,
+        importedRoot,
+        homeDir,
+        runsFilePath: path.join(homeDir, '.claude', 'state', 'no-such-runs.jsonl'),
+        now,
+        warnThreshold: 0.1,
+      });
+
+      const summary = result.data.summary;
+      assert.ok(summary.unmeasured_skills > 0, 'expected unmeasured skills');
+      assert.ok(
+        result.text.includes('WARNING'),
+        'expected an explicit warning when skills have no telemetry'
+      );
+      assert.match(result.text, /unmeasured/);
+      assert.strictEqual(
+        summary.healthy_skills,
+        0,
+        'skills without runs must not be counted as healthy'
+      );
+    })) passed++; else failed++;
+
+    // Isolated roots so the full-coverage case is guaranteed rather than
+    // conditional: every discovered skill needs a run inside the 30-day window.
+    if (test('renderDashboard omits the warning once every skill is measured', () => {
+      const coveredRoot = createTempDir('ecc-dashboard-covered-');
+      const coveredLearned = createTempDir('ecc-dashboard-covered-');
+      const coveredImported = createTempDir('ecc-dashboard-covered-');
+      const coveredRuns = path.join(homeDir, '.claude', 'state', 'covered-skill-runs.jsonl');
+      try {
+        for (const name of ['covered-a', 'covered-b']) {
+          createSkill(coveredRoot, name, `# ${name}\n`);
+        }
+        appendJsonl(coveredRuns, ['covered-a', 'covered-b'].map(skill_id => ({
+          skill_id,
+          skill_version: 'v1',
+          task_description: `Skill invocation: ${skill_id}`,
+          outcome: 'success',
+          recorded_at: now,
+        })));
+
+        const result = dashboard.renderDashboard({
+          skillsRoot: coveredRoot,
+          learnedRoot: coveredLearned,
+          importedRoot: coveredImported,
+          homeDir,
+          runsFilePath: coveredRuns,
+          now,
+          warnThreshold: 0.1,
+        });
+
+        assert.strictEqual(result.data.summary.unmeasured_skills, 0);
+        assert.strictEqual(result.data.summary.measured_skills, 2);
+        assert.ok(
+          !result.text.includes('WARNING'),
+          'no warning expected when every skill has a run in the window'
+        );
+        assert.strictEqual(
+          result.data.summary.total_skills,
+          result.data.summary.measured_skills + result.data.summary.unmeasured_skills,
+          'measured and unmeasured must partition the total'
+        );
+      } finally {
+        cleanupTempDir(coveredRoot);
+        cleanupTempDir(coveredLearned);
+        cleanupTempDir(coveredImported);
+      }
+    })) passed++; else failed++;
+
+    // The 30-day window can empty out while runs are still retained. Saying
+    // "nothing has ever been recorded" then is factually wrong.
+    if (test('the unmeasured warning distinguishes aged-out coverage from no history', () => {
+      const agedRuns = path.join(homeDir, '.claude', 'state', 'aged-skill-runs.jsonl');
+      const agedOut = new Date(Date.parse(now) - (60 * 24 * 60 * 60 * 1000)).toISOString();
+      appendJsonl(agedRuns, [{
+        skill_id: 'alpha',
+        skill_version: 'v1',
+        task_description: 'Skill invocation: alpha',
+        outcome: 'failure',
+        recorded_at: agedOut,
+      }]);
+
+      const result = dashboard.renderDashboard({
+        skillsRoot,
+        learnedRoot,
+        importedRoot,
+        homeDir,
+        runsFilePath: agedRuns,
+        now,
+        warnThreshold: 0.1,
+      });
+
+      assert.ok(result.data.summary.unmeasured_skills > 0, 'expected unmeasured skills');
+      assert.ok(result.text.includes('WARNING'), 'expected the warning');
+      assert.ok(
+        result.text.includes('aged out'),
+        'retained history outside the window must be described as aged out'
+      );
+      assert.ok(
+        !result.text.includes('Nothing has ever been recorded'),
+        'must not claim there is no history when runs are retained'
+      );
+    })) passed++; else failed++;
+
+    // Mixed coverage is the common case: some skills used recently, others not.
+// It must not be described as aged out, nor as having no history at all.
+    if (test('the unmeasured warning describes partial coverage as partial', () => {
+      const mixedRuns = path.join(homeDir, '.claude', 'state', 'mixed-skill-runs.jsonl');
+      appendJsonl(mixedRuns, [
+        {
+          skill_id: 'alpha',
+          skill_version: 'v1',
+          task_description: 'Skill invocation: alpha',
+          outcome: 'success',
+          recorded_at: now,
+        },
+        {
+          skill_id: 'beta',
+          skill_version: 'v1',
+          task_description: 'Skill invocation: beta',
+          outcome: 'success',
+          recorded_at: new Date(Date.parse(now) - (60 * 24 * 60 * 60 * 1000)).toISOString(),
+        },
+      ]);
+
+      const result = dashboard.renderDashboard({
+        skillsRoot,
+        learnedRoot,
+        importedRoot,
+        homeDir,
+        runsFilePath: mixedRuns,
+        now,
+        warnThreshold: 0.1,
+      });
+
+      assert.ok(result.text.includes('WARNING'), 'expected the warning');
+      assert.ok(
+        result.text.includes('Coverage is partial'),
+        'mixed recent and aged runs must be reported as partial coverage'
+      );
+      assert.ok(
+        !result.text.includes('aged out'),
+        'coverage is not aged out while recent runs exist'
+      );
+      assert.ok(
+        !result.text.includes('Nothing has ever been recorded'),
+        'runs are retained, so history is not missing'
+      );
+    })) passed++; else failed++;
+
+    // With no runs at all the stronger statement is the accurate one.
+    if (test('the unmeasured warning says nothing was ever recorded when there is no history', () => {
+      const result = dashboard.renderDashboard({
+        skillsRoot,
+        learnedRoot,
+        importedRoot,
+        homeDir,
+        runsFilePath: path.join(homeDir, '.claude', 'state', 'definitely-absent.jsonl'),
+        now,
+        warnThreshold: 0.1,
+      });
+
+      assert.ok(
+        result.text.includes('Nothing has ever been recorded'),
+        'an empty record set must be reported as missing telemetry'
+      );
+    })) passed++; else failed++;
+
     console.log('\nCLI integration:');
 
     if (test('CLI --dashboard --json returns valid JSON with all panels', () => {
diff --git a/tests/lib/skill-evolution.test.js b/tests/lib/skill-evolution.test.js
index bfff437bf..f2dd4e6ec 100644
--- a/tests/lib/skill-evolution.test.js
+++ b/tests/lib/skill-evolution.test.js
@@ -559,19 +559,67 @@ function runTests() {
       const summary = health.summarizeHealthReport(report);
       assert.deepStrictEqual(summary, {
         total_skills: 6,
-        healthy_skills: 5,
+        measured_skills: 2,
+        healthy_skills: 1,
         declining_skills: 1,
+        unmeasured_skills: 4,
       });
 
       const human = health.formatHealthReport(report, { json: false });
       assert.match(human, /alpha/);
       assert.match(human, /worsening/);
-      assert.match(
-        human,
-        new RegExp(`Skills: ${summary.total_skills} total, ${summary.healthy_skills} healthy, ${summary.declining_skills} declining`)
+      // Built from values rather than interpolated into a RegExp: the numbers are
+      // internally generated, but a literal comparison needs no pattern engine.
+      assert.ok(
+        human.includes(
+          `Skills: ${summary.total_skills} total, ${summary.measured_skills} measured, ` +
+          `${summary.healthy_skills} healthy, ${summary.declining_skills} declining, ` +
+          `${summary.unmeasured_skills} unmeasured`
+        ),
+        'summary line must report measured and unmeasured counts'
       );
     })) passed++; else failed++;
 
+    if (test('does not report unmeasured skills as healthy (#2463)', () => {
+      // A runs file that does not exist: skills are still discovered, so this
+      // is exactly the "discovered but never executed" state.
+      const emptyReport = health.collectSkillHealth({
+        repoRoot,
+        homeDir,
+        runsFilePath: path.join(homeDir, '.claude', 'state', 'no-such-runs.jsonl'),
+        now,
+        warnThreshold: 0.1,
+      });
+
+      const summary = health.summarizeHealthReport(emptyReport);
+      assert.ok(summary.total_skills > 0, 'expected discovered skills');
+      assert.strictEqual(summary.total_skills, summary.unmeasured_skills);
+      assert.strictEqual(summary.measured_skills, 0);
+      assert.strictEqual(summary.declining_skills, 0);
+      assert.strictEqual(
+        summary.healthy_skills,
+        0,
+        'a skill with no runs is unmeasured, never healthy'
+      );
+
+      // Healthy + declining must account for every measured skill, so a partial
+      // coverage state cannot over-report.
+      const partial = health.summarizeHealthReport({
+        skills: [
+          { skill_id: 'a', run_count_30d: 4, declining: false },
+          { skill_id: 'b', run_count_30d: 2, declining: true },
+          { skill_id: 'c', run_count_30d: 0, declining: false },
+        ],
+      });
+      assert.deepStrictEqual(partial, {
+        total_skills: 3,
+        measured_skills: 2,
+        healthy_skills: 1,
+        declining_skills: 1,
+        unmeasured_skills: 1,
+      });
+    })) passed++; else failed++;
+
     if (test('treats an unsnapshotted SKILL.md as v1 and orders last_run by actual time', () => {
       const gammaSkillDir = createSkill(skillsRoot, 'gamma', '# Gamma v1\n');
       const offsetRunsFile = path.join(homeDir, '.claude', 'state', 'offset-skill-runs.jsonl');
diff --git a/tests/opencode-plugin-hooks.test.js b/tests/opencode-plugin-hooks.test.js
index a6c3ed2ae..e327f6ebf 100644
--- a/tests/opencode-plugin-hooks.test.js
+++ b/tests/opencode-plugin-hooks.test.js
@@ -79,6 +79,90 @@ async function withTempProject(files, fn) {
   }
 }
 
+/**
+ * Build a fake ECC installation root and copy the compiled plugin into it.
+ *
+ * Reproduces a *modular* install: the runtime scripts the plugin needs are
+ * present, but `skills/` is absent because that directory belongs to the optional
+ * `workflow-quality` module. `options.withSkills` adds it back so the same helper
+ * can prove the difference is what matters.
+ *
+ * @param {{withSkills?: boolean}} [options]
+ * @returns {string} path to the fake installation root
+ */
+function createModularInstallRoot(options = {}) {
+  const root = fs.mkdtempSync(path.join(os.tmpdir(), "ecc-opencode-install-"))
+  const distSource = path.join(__dirname, "..", ".opencode", "dist")
+  const repoRoot = path.join(__dirname, "..")
+
+  for (const dir of ["plugins", "tools"]) {
+    fs.cpSync(path.join(distSource, dir), path.join(root, dir), { recursive: true })
+  }
+
+  // The runtime files an install without `workflow-quality` still ships. Copied
+  // whole rather than file-by-file: scripts/lib has transitive requires (utils.js
+  // alone pulls in agent-data-home), and a partial copy would fail the import for
+  // reasons unrelated to what this test is about.
+  for (const dir of ["hooks", "lib"]) {
+    fs.cpSync(path.join(repoRoot, "scripts", dir), path.join(root, "scripts", dir), {
+      recursive: true,
+    })
+  }
+
+  fs.writeFileSync(
+    path.join(root, "package.json"),
+    JSON.stringify({ name: "ecc-universal", version: "0.0.0-test" })
+  )
+
+  // The plugin's tools import @opencode-ai/plugin, which resolves through
+  // node_modules — a real installation has it, an isolated temp dir would not.
+  fs.symlinkSync(path.join(repoRoot, "node_modules"), path.join(root, "node_modules"), "junction")
+
+  if (options.withSkills) {
+    fs.mkdirSync(path.join(root, "skills", "continuous-learning-v2"), { recursive: true })
+    fs.writeFileSync(path.join(root, "skills", "continuous-learning-v2", "SKILL.md"), "# sentinel\n")
+  }
+
+  return root
+}
+
+/** Load the compiled plugin from a fake install root instead of the repo dist. */
+async function loadPluginFrom(root) {
+  return import(pathToFileURL(path.join(root, "plugins", "ecc-hooks.js")).href)
+}
+
+/**
+ * Restore environment variables saved with Object.fromEntries of name -> value.
+ *
+ * @param {Record<string, string | undefined>} saved
+ * @returns {void}
+ */
+function restoreEnv(saved) {
+  for (const [key, value] of Object.entries(saved)) {
+    if (value === undefined) delete process.env[key]
+    else process.env[key] = value
+  }
+}
+
+/**
+ * Read the skill-runs sink the tracker wrote under `homeDir`.
+ *
+ * The tracker resolves the sink from the *current* home, so the telemetry tests
+ * move HOME/USERPROFILE to a temp dir and read from there.
+ *
+ * @param {string} homeDir
+ * @returns {Array<Record<string, unknown>>}
+ */
+function readRecordedRuns(homeDir) {
+  const runsFilePath = path.join(homeDir, ".claude", "state", "skill-runs.jsonl")
+  if (!fs.existsSync(runsFilePath)) return []
+  return fs
+    .readFileSync(runsFilePath, "utf8")
+    .split("\n")
+    .filter(Boolean)
+    .map((line) => JSON.parse(line))
+}
+
 async function main() {
   console.log("\n=== Testing OpenCode plugin hooks ===\n")
 
@@ -381,6 +465,232 @@ async function main() {
         }
       ),
     ],
+    [
+      "skill telemetry records a run through the shared tracker (#2463)",
+      async () => withTempProject([], async (projectDir) => {
+        const homeDir = fs.mkdtempSync(path.join(os.tmpdir(), "ecc-opencode-skill-home-"))
+        const previousHome = process.env.HOME
+        const previousUserProfile = process.env.USERPROFILE
+        process.env.HOME = homeDir
+        process.env.USERPROFILE = homeDir
+        try {
+          const client = createClient()
+          const $ = createFailingShell()
+          const hooks = await ECCHooksPlugin({ client, $, directory: projectDir })
+
+          assert.ok(
+            !client.logs.some(
+              (entry) =>
+                entry.level === "warn" && entry.message.includes("skill-run telemetry disabled")
+            ),
+            "Did not expect telemetry to be disabled when scripts/hooks/skill-run-tracker.js is present"
+          )
+
+          // OpenCode names the tool `skill`; the id arrives on `name`.
+          await hooks["tool.execute.after"]({ tool: "skill", args: { name: "code-review" } }, {})
+
+          const records = readRecordedRuns(homeDir)
+
+          assert.strictEqual(records.length, 1, "Expected exactly one recorded skill run")
+          assert.strictEqual(records[0].skill_id, "code-review")
+          assert.strictEqual(records[0].outcome, "success")
+        } finally {
+          restoreEnv({ HOME: previousHome, USERPROFILE: previousUserProfile })
+          fs.rmSync(homeDir, { recursive: true, force: true })
+        }
+      }),
+    ],
+    [
+      "skill telemetry records nothing for other tools",
+      async () => withTempProject([], async (projectDir) => {
+        const homeDir = fs.mkdtempSync(path.join(os.tmpdir(), "ecc-opencode-skill-home-"))
+        const previousHome = process.env.HOME
+        const previousUserProfile = process.env.USERPROFILE
+        process.env.HOME = homeDir
+        process.env.USERPROFILE = homeDir
+        try {
+          const client = createClient()
+          const $ = createFailingShell()
+          const hooks = await ECCHooksPlugin({ client, $, directory: projectDir })
+
+          await hooks["tool.execute.after"]({ tool: "edit", args: { path: "src/a.ts" } }, {})
+          await hooks["tool.execute.after"]({ tool: "bash", args: { command: "ls" } }, {})
+
+          assert.strictEqual(readRecordedRuns(homeDir).length, 0)
+        } finally {
+          restoreEnv({ HOME: previousHome, USERPROFILE: previousUserProfile })
+          fs.rmSync(homeDir, { recursive: true, force: true })
+        }
+      }),
+    ],
+    [
+      "skill telemetry respects the minimal hook profile",
+      async () => withTempProject([], async (projectDir) => {
+        const homeDir = fs.mkdtempSync(path.join(os.tmpdir(), "ecc-opencode-skill-home-"))
+        const previousHome = process.env.HOME
+        const previousUserProfile = process.env.USERPROFILE
+        const previousProfile = process.env.ECC_HOOK_PROFILE
+        process.env.HOME = homeDir
+        process.env.USERPROFILE = homeDir
+        process.env.ECC_HOOK_PROFILE = "minimal"
+        try {
+          const client = createClient()
+          const $ = createFailingShell()
+          const hooks = await ECCHooksPlugin({ client, $, directory: projectDir })
+
+          await hooks["tool.execute.after"]({ tool: "skill", args: { name: "code-review" } }, {})
+
+          assert.strictEqual(readRecordedRuns(homeDir).length, 0)
+        } finally {
+          restoreEnv({
+            HOME: previousHome,
+            USERPROFILE: previousUserProfile,
+            ECC_HOOK_PROFILE: previousProfile,
+          })
+          fs.rmSync(homeDir, { recursive: true, force: true })
+        }
+      }),
+    ],
+    [
+      "shell.env exports CLAUDE_PLUGIN_ROOT so commands can resolve the ECC root",
+      async () => withTempProject([], async (projectDir) => {
+        const previousRoot = process.env.CLAUDE_PLUGIN_ROOT
+        delete process.env.CLAUDE_PLUGIN_ROOT
+        try {
+          const client = createClient()
+          const $ = createFailingShell()
+          const hooks = await ECCHooksPlugin({ client, $, directory: projectDir })
+
+          const output = { env: {} }
+          await hooks["shell.env"]({ cwd: projectDir }, output)
+
+          assert.deepStrictEqual($.calls, [], `Unexpected shell probes: ${$.calls.join(", ")}`)
+          assert.ok(output.env.CLAUDE_PLUGIN_ROOT, "Expected CLAUDE_PLUGIN_ROOT to be exported")
+
+          // The resolver consumes this variable verbatim, so it must be a
+          // directory that actually carries the ECC script tree.
+          const root = output.env.CLAUDE_PLUGIN_ROOT
+          assert.ok(
+            fs.existsSync(path.join(root, "scripts", "lib", "resolve-ecc-root.js")),
+            `Expected ${root} to contain scripts/lib/resolve-ecc-root.js`
+          )
+        } finally {
+          restoreEnv({ CLAUDE_PLUGIN_ROOT: previousRoot })
+        }
+      }),
+    ],
+    [
+      "shell.env does not override a CLAUDE_PLUGIN_ROOT the user set",
+      async () => withTempProject([], async (projectDir) => {
+        const explicitRoot = path.join(projectDir, "explicit-root")
+        const previousRoot = process.env.CLAUDE_PLUGIN_ROOT
+        process.env.CLAUDE_PLUGIN_ROOT = explicitRoot
+        try {
+          const client = createClient()
+          const $ = createFailingShell()
+          const hooks = await ECCHooksPlugin({ client, $, directory: projectDir })
+
+          const output = { env: {} }
+          await hooks["shell.env"]({ cwd: projectDir }, output)
+
+          assert.strictEqual(
+            output.env.CLAUDE_PLUGIN_ROOT,
+            undefined,
+            "the plugin must not inject its own root over one the user set"
+          )
+          assert.strictEqual(
+            process.env.CLAUDE_PLUGIN_ROOT,
+            explicitRoot,
+            "the user's value must survive untouched"
+          )
+        } finally {
+          restoreEnv({ CLAUDE_PLUGIN_ROOT: previousRoot })
+        }
+      }),
+    ],
+    [
+      "shell.env does not clobber a root another plugin already put in output.env",
+      async () => withTempProject([], async (projectDir) => {
+        const previousRoot = process.env.CLAUDE_PLUGIN_ROOT
+        delete process.env.CLAUDE_PLUGIN_ROOT
+        try {
+          const client = createClient()
+          const $ = createFailingShell()
+          const hooks = await ECCHooksPlugin({ client, $, directory: projectDir })
+
+          // An earlier plugin chose a root. The final merge below spreads our env
+          // over output.env, so a guard that only checked process.env would
+          // replace this value.
+          const output = { env: { CLAUDE_PLUGIN_ROOT: path.join(projectDir, "other-plugin-root") } }
+          await hooks["shell.env"]({ cwd: projectDir }, output)
+
+          assert.strictEqual(
+            output.env.CLAUDE_PLUGIN_ROOT,
+            path.join(projectDir, "other-plugin-root"),
+            "an existing output.env root must be preserved"
+          )
+        } finally {
+          restoreEnv({ CLAUDE_PLUGIN_ROOT: previousRoot })
+        }
+      }),
+    ],
+    [
+      "telemetry and root export work in a modular install without optional skills",
+      async () => {
+        // Regression: root detection probed skills/continuous-learning-v2, which
+        // belongs to the optional workflow-quality module. An install with
+        // platform-configs + hooks-runtime but without workflow-quality still
+        // ships the plugin and the tracker, and it lost both fixes.
+        const root = createModularInstallRoot()
+        const projectDir = fs.mkdtempSync(path.join(os.tmpdir(), "ecc-opencode-modular-"))
+        const homeDir = fs.mkdtempSync(path.join(os.tmpdir(), "ecc-opencode-skill-home-"))
+        const previousHome = process.env.HOME
+        const previousUserProfile = process.env.USERPROFILE
+        const previousRoot = process.env.CLAUDE_PLUGIN_ROOT
+        process.env.HOME = homeDir
+        process.env.USERPROFILE = homeDir
+        delete process.env.CLAUDE_PLUGIN_ROOT
+        try {
+          assert.ok(
+            !fs.existsSync(path.join(root, "skills")),
+            "the fixture must not contain skills/, or it would not test the modular case"
+          )
+
+          const { ECCHooksPlugin: modularPlugin } = await loadPluginFrom(root)
+          const client = createClient()
+          const hooks = await modularPlugin({
+            client,
+            $: createFailingShell(),
+            directory: projectDir,
+          })
+
+          assert.ok(
+            !client.logs.some(
+              (entry) =>
+                entry.level === "warn" && entry.message.includes("skill-run telemetry disabled")
+            ),
+            "telemetry must stay enabled when the optional skills module is absent"
+          )
+
+          await hooks["tool.execute.after"]({ tool: "skill", args: { name: "code-review" } }, {})
+          assert.strictEqual(readRecordedRuns(homeDir).length, 1, "expected the run to be recorded")
+
+          const output = { env: {} }
+          await hooks["shell.env"]({ cwd: projectDir }, output)
+          assert.strictEqual(output.env.CLAUDE_PLUGIN_ROOT, root)
+          assert.strictEqual(output.env.ECC_VERSION, "0.0.0-test")
+        } finally {
+          restoreEnv({
+            HOME: previousHome,
+            USERPROFILE: previousUserProfile,
+            CLAUDE_PLUGIN_ROOT: previousRoot,
+          })
+          fs.rmSync(root, { recursive: true, force: true })
+          fs.rmSync(projectDir, { recursive: true, force: true })
+          fs.rmSync(homeDir, { recursive: true, force: true })
+        }
+      },
+    ],
   ]
 
   let passed = 0