@hiveai/cli 0.12.0 → 0.12.3

package/README.md

@@ -187,6 +187,35 @@ Autogenerated sensors are conservative: they start as `warn` and `autogen: true`
 high-confidence sensors to `severity: block`, which makes a deterministic pre-commit blocker when the
 sensor matches added diff lines.
 
+### `haive eval`
+
+Run the repeatable quality gate for hAIve itself or for a project using hAIve:
+
+```bash
+haive eval
+haive eval --semantic-only
+haive eval --spec .ai/eval/spec.json --fail-under 80
+```
+
+Without `--spec`, hAIve synthesizes retrieval cases from anchored memories. If `.ai/eval/spec.json`
+exists, it is loaded automatically and merged with those synthesized retrieval cases. Use that file
+for labeled sensor cases and hard retrieval probes so CI measures both “did the right memory surface?”
+and “did the executable guardrail fire?”.
+
+### `haive doctor`
+
+`doctor` is the first stop when hAIve feels inconsistent locally:
+
+```bash
+haive doctor
+haive doctor --json
+haive doctor --fix
+```
+
+It reports missing `pnpm`, stale workspace `dist` artifacts after a pull, global CLI/MCP version skew,
+outdated code-search indexes, memory-lint findings, and harness coverage. The output is intentionally
+actionable: every setup finding should carry the exact command to run next.
+
 ### `haive benchmark`
 
 Turn hAIve-vs-plain agent trials into a repeatable demo/report.

package/dist/index.js

@@ -2743,7 +2743,7 @@ ${SEED_FOOTER(stack)}` });
 }
 
 // src/commands/init.ts
-var HAIVE_GITHUB_ACTION_REF = `v${"0.12.0"}`;
+var HAIVE_GITHUB_ACTION_REF = `v${"0.12.3"}`;
 var PROJECT_CONTEXT_TEMPLATE = `# Project context
 
 > Generated by \`haive init\`. Run \`haive init --bootstrap\` to auto-fill from your codebase,
@@ -3782,6 +3782,7 @@ import {
   loadMemoriesFromDir as loadMemoriesFromDir14,
   loadUsageIndex as loadUsageIndex8,
   memoryMatchesAnchorPaths as memoryMatchesAnchorPaths22,
+  rankMemoriesLexical as rankMemoriesLexical2,
   queryCodeMap as queryCodeMap2,
   resolveBriefingBudget as resolveBriefingBudget2,
   serializeMemory as serializeMemory9,
@@ -5472,13 +5473,25 @@ async function getBriefing(input, ctx) {
       }
       if (act.applicable && act.activated) activatedSkills.add(id);
     }
+    const lexNorm = /* @__PURE__ */ new Map();
+    if (input.task) {
+      const candidates = [...seen.keys()].map((id) => byId.get(id)).filter((x) => Boolean(x));
+      const lex = rankMemoriesLexical2(candidates, input.task, candidates.length);
+      const maxScore = lex.scores.reduce((m, s) => s > m ? s : m, 0);
+      if (maxScore > 0) {
+        lex.ranked.forEach((loaded, i) => {
+          lexNorm.set(loaded.memory.frontmatter.id, (lex.scores[i] ?? 0) / maxScore);
+        });
+      }
+    }
     const ranked = [...seen.values()].sort((a, b) => {
       const reasonScore = (m) => (m.type === "attempt" ? 3 : 0) + (m.reasons.includes("anchor") ? 4 : 0) + (m.reasons.includes("symbol") ? 4 : 0) + (m.reasons.includes("module") ? 2 : 0) + (m.reasons.includes("semantic") ? 2 : 0) + (m.reasons.includes("domain") ? 1 : 0);
       const confidenceScore = (m) => m.confidence === "authoritative" ? 4 : m.confidence === "trusted" ? 3 : m.confidence === "low" ? 1 : m.confidence === "stale" ? -2 : 0;
       const impactScore = (m) => (m.impact_score ?? 0) * 3;
       const activationBoost = (m) => activatedSkills.has(m.id) ? 5 : 0;
-      const sa = priorityRank(classifyMemoryPriority(a, byId.get(a.id), input.files, input.symbols)) * 100 + reasonScore(a) + confidenceScore(a) + impactScore(a) + activationBoost(a) + (a.semantic_score ?? 0);
-      const sb = priorityRank(classifyMemoryPriority(b, byId.get(b.id), input.files, input.symbols)) * 100 + reasonScore(b) + confidenceScore(b) + impactScore(b) + activationBoost(b) + (b.semantic_score ?? 0);
+      const lexScore = (m) => 12 * (lexNorm.get(m.id) ?? 0);
+      const sa = priorityRank(classifyMemoryPriority(a, byId.get(a.id), input.files, input.symbols)) * 100 + reasonScore(a) + confidenceScore(a) + impactScore(a) + activationBoost(a) + lexScore(a) + (a.semantic_score ?? 0);
+      const sb = priorityRank(classifyMemoryPriority(b, byId.get(b.id), input.files, input.symbols)) * 100 + reasonScore(b) + confidenceScore(b) + impactScore(b) + activationBoost(b) + lexScore(b) + (b.semantic_score ?? 0);
       return sb - sa;
     });
     for (const mem of ranked.slice(0, briefingMaxMemories)) {
@@ -7518,7 +7531,7 @@ When done, respond with: "Imported N memories: [list of IDs]" or "Nothing action
   };
 }
 var SERVER_NAME = "haive";
-var SERVER_VERSION = "0.12.0";
+var SERVER_VERSION = "0.12.3";
 function jsonResult(data) {
   return {
     content: [
@@ -12246,7 +12259,8 @@ function registerEval(program2) {
     }
     const k = Math.max(1, parseInt(opts.top ?? "8", 10) || 8);
     const ctx = { paths };
-    const spec = await resolveSpec(opts, paths.memoriesDir);
+    const resolvedSpec = await resolveSpec(opts, root, paths.memoriesDir);
+    const spec = resolvedSpec.spec;
     if ((spec.retrieval?.length ?? 0) === 0 && (spec.sensors?.length ?? 0) === 0) {
       ui.warn("No eval cases (no anchored memories and no --spec). Nothing to score.");
       return;
@@ -12271,9 +12285,9 @@ function registerEval(program2) {
     }
     const report = buildReport(retrievalAgg, sensorAgg);
     if (opts.json) {
-      console.log(JSON.stringify({ root, k, report }, null, 2));
+      console.log(JSON.stringify({ root, k, spec_source: resolvedSpec.source, report }, null, 2));
     } else {
-      const md = renderMarkdown2(root, k, report);
+      const md = renderMarkdown2(root, k, resolvedSpec.source, report);
       if (opts.out) {
         const outFile = path43.isAbsolute(opts.out) ? opts.out : path43.join(root, opts.out);
         await writeFile29(outFile, md, "utf8");
@@ -12294,14 +12308,31 @@ function registerEval(program2) {
     }
   });
 }
-async function resolveSpec(opts, memoriesDir) {
+async function resolveSpec(opts, root, memoriesDir) {
   if (opts.spec) {
     const file = path43.resolve(opts.spec);
     const raw = await readFile20(file, "utf8");
-    return JSON.parse(raw);
+    return { spec: JSON.parse(raw), source: file };
+  }
+  const defaultSpec = path43.join(root, ".ai", "eval", "spec.json");
+  if (existsSync64(defaultSpec)) {
+    const raw = await readFile20(defaultSpec, "utf8");
+    const explicit = JSON.parse(raw);
+    const memories2 = await loadMemoriesFromDir26(memoriesDir);
+    const synthesized = synthesizeSelfEvalCases(memories2, { includeFiles: !opts.semanticOnly });
+    return {
+      spec: {
+        retrieval: [...synthesized, ...explicit.retrieval ?? []],
+        sensors: explicit.sensors ?? []
+      },
+      source: ".ai/eval/spec.json + synthesized anchored retrieval"
+    };
   }
   const memories = await loadMemoriesFromDir26(memoriesDir);
-  return { retrieval: synthesizeSelfEvalCases(memories, { includeFiles: !opts.semanticOnly }) };
+  return {
+    spec: { retrieval: synthesizeSelfEvalCases(memories, { includeFiles: !opts.semanticOnly }) },
+    source: "synthesized anchored retrieval"
+  };
 }
 async function runRetrieval(c, k, ctx) {
   const out = await getBriefing(
@@ -12333,11 +12364,12 @@ async function runSensorCase(c, ctx) {
 function pct(n) {
   return `${Math.round(n * 100)}%`;
 }
-function renderMarkdown2(root, k, report) {
+function renderMarkdown2(root, k, source, report) {
   const lines = [
     "# hAIve eval report",
     "",
     `Project: \`${root}\` \xB7 top-k: ${k}`,
+    `Spec: ${source}`,
     "",
     `## Overall score: ${report.score}/100`,
     ""
@@ -12969,14 +13001,15 @@ function registerDoctor(program2) {
         fix: "Edit .ai/haive.config.json: set autoSessionEnd: true (or re-run `haive init` without --manual)."
       });
     }
-    findings.push(...await collectInstallFindings(root, "0.12.0"));
+    findings.push(...await collectInstallFindings(root, "0.12.3"));
+    findings.push(...await collectToolchainFindings(root));
     try {
       const legacyRaw = execSync3("haive-mcp --version", {
         encoding: "utf8",
         timeout: 3e3,
         stdio: ["ignore", "pipe", "ignore"]
       }).trim();
-      const cliVersion = "0.12.0";
+      const cliVersion = "0.12.3";
       if (legacyRaw && legacyRaw !== cliVersion) {
         findings.push({
           severity: "warn",
@@ -13253,6 +13286,7 @@ function isSearchTool(name) {
 async function collectInstallFindings(root, expectedVersion) {
   const findings = [];
   findings.push(...await collectWorkspaceVersionFindings(root, expectedVersion));
+  findings.push(...await collectDistFreshnessFindings(root, expectedVersion));
   const haiveBins = listHaiveBins();
   if (haiveBins.length === 0) {
     findings.push({
@@ -13316,6 +13350,71 @@ which -a haive`
   }
   return findings;
 }
+async function collectToolchainFindings(root) {
+  const findings = [];
+  const pkg = await readJson(
+    path46.join(root, "package.json")
+  );
+  const wantsPnpm = pkg?.packageManager?.startsWith("pnpm@") || Object.values(pkg?.scripts ?? {}).some((script) => /\bpnpm\b/.test(script));
+  if (!wantsPnpm) return findings;
+  const expected = pkg?.packageManager?.replace(/^pnpm@/, "") ?? "9.14.2";
+  if (!commandExists2("pnpm", ["--version"])) {
+    const corepackAvailable = commandExists2("corepack", ["--version"]);
+    findings.push({
+      severity: "warn",
+      code: "pnpm-not-on-path",
+      message: `This workspace uses pnpm${expected ? ` ${expected}` : ""}, but no pnpm binary is available on PATH. Local build/test commands may fail even though CI works.`,
+      fix: corepackAvailable ? `corepack prepare pnpm@${expected} --activate` : `npx pnpm@${expected} install --frozen-lockfile`,
+      section: "Agent coverage"
+    });
+  }
+  return findings;
+}
+async function collectDistFreshnessFindings(root, expectedVersion) {
+  const findings = [];
+  const isHaiveWorkspace = (await readJson(path46.join(root, "package.json")))?.name === "haive-monorepo";
+  if (!isHaiveWorkspace) return findings;
+  const cliDist = path46.join(root, "packages/cli/dist/index.js");
+  if (!existsSync67(cliDist)) {
+    findings.push({
+      severity: "warn",
+      code: "workspace-dist-missing",
+      message: "packages/cli/dist/index.js is missing; local hAIve smoke commands cannot reflect source changes.",
+      fix: "pnpm -r build",
+      section: "Agent coverage"
+    });
+    return findings;
+  }
+  const distVersion = versionForNodeEntrypoint(cliDist);
+  if (distVersion && distVersion !== expectedVersion) {
+    findings.push({
+      severity: "warn",
+      code: "workspace-dist-version-mismatch",
+      message: `packages/cli/dist/index.js reports ${distVersion}, but this source build expects ${expectedVersion}. Run a fresh workspace build after pull before trusting local doctor/enforce output.`,
+      fix: "pnpm -r build\npnpm check:artifacts",
+      section: "Agent coverage"
+    });
+  }
+  const sourceFiles = [
+    "packages/core/src/index.ts",
+    "packages/mcp/src/server.ts",
+    "packages/cli/src/index.ts"
+  ].map((rel) => path46.join(root, rel)).filter(existsSync67);
+  if (sourceFiles.length > 0) {
+    const distMtime = statSync2(cliDist).mtimeMs;
+    const newestSource = Math.max(...sourceFiles.map((file) => statSync2(file).mtimeMs));
+    if (newestSource > distMtime + 1e3) {
+      findings.push({
+        severity: "info",
+        code: "workspace-dist-older-than-source",
+        message: "Built CLI artifacts are older than key source files; rebuild before running release smoke checks.",
+        fix: "pnpm -r build\npnpm check:artifacts",
+        section: "Agent coverage"
+      });
+    }
+  }
+  return findings;
+}
 async function collectWorkspaceVersionFindings(root, expectedVersion) {
   const findings = [];
   const rootPkg = await readJson(path46.join(root, "package.json"));
@@ -13419,6 +13518,30 @@ function versionForBinary(bin) {
     return null;
   }
 }
+function versionForNodeEntrypoint(file) {
+  try {
+    const out = execFileSync(process.execPath, [file, "--version"], {
+      encoding: "utf8",
+      timeout: 3e3,
+      stdio: ["ignore", "pipe", "ignore"]
+    }).trim();
+    return out.match(/\d+\.\d+\.\d+/)?.[0] ?? null;
+  } catch {
+    return null;
+  }
+}
+function commandExists2(command, args) {
+  try {
+    execFileSync(command, args, {
+      encoding: "utf8",
+      timeout: 3e3,
+      stdio: ["ignore", "pipe", "ignore"]
+    });
+    return true;
+  } catch {
+    return false;
+  }
+}
 function extractAbsoluteHaiveBins(text) {
   const out = /* @__PURE__ */ new Set();
   const re = /(["'\s])((?:\/[^"'\s]+)*\/haive)\b/g;
@@ -14477,7 +14600,7 @@ async function buildEnforcementReport(dir, stage, sessionId) {
       findings: [{ severity: "info", code: "enforcement-off", message: "hAIve enforcement is disabled." }]
     });
   }
-  findings.push(...await inspectIntegrationVersions(root, "0.12.0"));
+  findings.push(...await inspectIntegrationVersions(root, "0.12.3"));
   if (config.enforcement?.requireBriefingFirst !== false && stage !== "ci") {
     const hasBriefing = await hasRecentBriefingMarker2(paths, sessionId);
     findings.push(hasBriefing ? { severity: "ok", code: "briefing-loaded", message: "A recent hAIve briefing marker exists." } : {
@@ -14615,6 +14738,15 @@ async function verifyDecisionCoverage(paths, stage, sessionId) {
     }];
   }
   const marker = await readRecentBriefingMarker(paths, sessionId);
+  if (stage === "ci" && !marker) {
+    return [{
+      severity: "ok",
+      code: "decision-coverage-ci-pass",
+      message: `CI surfaced ${relevant.length} relevant anchored decision/polic${relevant.length === 1 ? "y" : "ies"} for ${changedFiles.length} changed file(s). Runtime briefing markers are local-only and are not expected on GitHub Actions.`,
+      memory_ids: relevant.slice(0, 20).map((memory2) => memory2.frontmatter.id),
+      affected_files: changedFiles.slice(0, 10)
+    }];
+  }
   const consulted = new Set(marker?.memory_ids ?? []);
   const missing = relevant.filter((memory2) => !consulted.has(memory2.frontmatter.id));
   if (missing.length === 0) {
@@ -15476,7 +15608,7 @@ function shellQuote(value) {
 
 // src/index.ts
 var program = new Command56();
-program.name("haive").description("hAIve - repo-native memory and context policy for coding-agent harnesses").version("0.12.0").option("--advanced", "show maintenance and experimental commands in help");
+program.name("haive").description("hAIve - repo-native memory and context policy for coding-agent harnesses").version("0.12.3").option("--advanced", "show maintenance and experimental commands in help");
 registerInit(program);
 registerWelcome(program);
 registerResolveProject(program);