muonroi-cli 1.4.0 → 1.4.1

package/src/product-loop/progress-snapshot.ts

@@ -1,216 +0,0 @@
-/**
- * src/product-loop/progress-snapshot.ts
- *
- * Pure projection: reads Backlog + SprintPlan + interaction_logs and returns
- * a ProgressSnapshot. No LLM, no external writes.
- *
- * Both /status and the P8 reporter consume this module.
- * renderSnapshotMarkdown() is a pure string function — no I/O — so tests and
- * the reporter can call it without a live sprint-runner.
- */
-
-import { getDatabase } from "../storage/db.js";
-import { readBacklog } from "./backlog-store.js";
-import { readSprintPlan } from "./sprint-store.js";
-import type { BacklogItemStatus, ProgressSnapshot, ProgressSnapshotBlocker, ProgressSnapshotItem } from "./types.js";
-
-export interface ComputeSnapshotInput {
-  flowDir: string;
-  runId: string;
-  productSlug: string;
-}
-
-// ─── interaction_logs read helper ────────────────────────────────────────────
-
-interface SprintStageRow {
-  metadata_json: string | null;
-  created_at: string;
-}
-
-/**
- * Read the latest sprint_stage interaction log row for a runId.
- * Queries via the DB directly (no read helper exists in storage/ for this query).
- * Returns null on any error or if no row exists.
- */
-function readLatestSprintStage(runId: string): { sprintIndex: number; stage: string; createdAt: string } | null {
-  try {
-    const db = getDatabase();
-    const row = db
-      .prepare(
-        `SELECT metadata_json, created_at
-         FROM interaction_logs
-         WHERE session_id = ? AND event_type = 'ui_interaction' AND event_subtype = 'sprint_stage'
-         ORDER BY id DESC
-         LIMIT 1`,
-      )
-      .get(runId) as SprintStageRow | undefined;
-
-    if (!row) return null;
-    const meta = row.metadata_json ? (JSON.parse(row.metadata_json) as Record<string, unknown>) : null;
-    if (!meta) return null;
-
-    const sprintIndex = typeof meta.sprintIndex === "number" ? meta.sprintIndex : null;
-    const stage = typeof meta.stage === "string" ? meta.stage : null;
-    if (sprintIndex === null || !stage) return null;
-
-    return { sprintIndex, stage, createdAt: row.created_at };
-  } catch {
-    return null;
-  }
-}
-
-function capitalize(s: string): string {
-  if (!s) return s;
-  return s.charAt(0).toUpperCase() + s.slice(1);
-}
-
-// ─── Main export ──────────────────────────────────────────────────────────────
-
-/**
- * Compute a ProgressSnapshot from disk + DB. Pure I/O — no LLM.
- */
-export async function computeProgressSnapshot(input: ComputeSnapshotInput): Promise<ProgressSnapshot> {
-  const { flowDir, runId, productSlug } = input;
-
-  // 1. Read Backlog
-  const backlog = await readBacklog(flowDir, runId);
-
-  // 2. Read SprintPlan
-  const sprintPlan = await readSprintPlan(flowDir, runId);
-
-  // 3. Read latest sprint stage from interaction_logs
-  const stageRow = readLatestSprintStage(runId);
-  const workerCurrentStage = stageRow ? `Sprint ${stageRow.sprintIndex} — ${capitalize(stageRow.stage)}` : null;
-  const workerLastEventUtc = stageRow ? stageRow.createdAt : null;
-
-  // 4. Clarify status — P5 persists clarified-spec.json; best-effort read.
-  // TODO: read clarified-spec.json from flowDir/runs/runId/ when P5 persists it to disk.
-  //       For now, assume ready=true with no gaps if backlog exists (backlog can only
-  //       be built after clarification is complete).
-  const clarifyReady = backlog !== null;
-  const clarifyGaps: string[] = [];
-
-  // 5. Backlog metrics
-  const backlogTotal = backlog?.items.length ?? 0;
-  const backlogV1Count = backlog?.items.filter((i) => i.mvp_priority === "v1").length ?? 0;
-  const backlogDeferredCount =
-    backlog?.items.filter((i) => i.mvp_priority === "v2" || i.mvp_priority === "later").length ?? 0;
-
-  // 6. Sprint metrics
-  const sprintTotal = sprintPlan?.sprints.length ?? 0;
-  const activeSprint = sprintPlan?.sprints.find((s) => s.id === sprintPlan.activeSprintId) ?? null;
-  const activeSprintNumber = activeSprint?.number ?? null;
-  const activeSprintGoal = activeSprint?.goal ?? null;
-
-  // 7. Active sprint items
-  const activeSprintItems: ProgressSnapshotItem[] = [];
-  if (activeSprint && backlog) {
-    for (const itemId of activeSprint.itemIds) {
-      const item = backlog.items.find((i) => i.id === itemId);
-      if (!item) continue;
-      activeSprintItems.push({
-        id: item.id,
-        title: item.title,
-        status: item.status,
-        // v1 heuristic: count done items as fully met; others as 0%.
-        // Per-criterion tracking is P8+ work.
-        criteriaMet: item.status === "done" ? item.acceptance_criteria.length : 0,
-        criteriaTotal: item.acceptance_criteria.length,
-      });
-    }
-  }
-
-  // 8. Percent done — count completed items / total in sprint × 100
-  const totalItems = activeSprintItems.length;
-  const doneItems = activeSprintItems.filter((i) => i.status === "done").length;
-  const activeSprintPercentDone = totalItems > 0 ? Math.round((doneItems / totalItems) * 1000) / 10 : 0;
-
-  // 9. Blockers: backlog items with status="blocked" in the active sprint
-  const blockers: ProgressSnapshotBlocker[] = [];
-  if (activeSprint && backlog) {
-    for (const itemId of activeSprint.itemIds) {
-      const item = backlog.items.find((i) => i.id === itemId);
-      if (!item || item.status !== "blocked") continue;
-      const reason = item.blockers && item.blockers.length > 0 ? `blocked by ${item.blockers[0]}` : "blocked";
-      blockers.push({ itemId: item.id, title: item.title, reason });
-    }
-  }
-
-  return {
-    runId,
-    productSlug,
-    capturedAtUtc: new Date().toISOString(),
-    clarifyReady,
-    clarifyGaps,
-    backlogTotal,
-    backlogV1Count,
-    backlogDeferredCount,
-    sprintTotal,
-    activeSprintNumber,
-    activeSprintGoal,
-    activeSprintPercentDone,
-    activeSprintItems,
-    blockers,
-    workerLastEventUtc,
-    workerCurrentStage,
-  };
-}
-
-// ─── Markdown renderer ────────────────────────────────────────────────────────
-
-function statusIcon(status: BacklogItemStatus): string {
-  if (status === "done") return "[x]";
-  if (status === "in_progress") return "[~]";
-  return "[ ]";
-}
-
-/**
- * Render a ProgressSnapshot as a markdown progress card.
- * Pure function — no I/O. Both /status and sprint-runner use this.
- */
-export function renderSnapshotMarkdown(snapshot: ProgressSnapshot): string {
-  const lines: string[] = [];
-
-  lines.push(`## Progress — ${snapshot.productSlug}`);
-  lines.push("");
-  lines.push(
-    `**Backlog:** ${snapshot.backlogTotal} total · ${snapshot.backlogV1Count} v1 · ${snapshot.backlogDeferredCount} deferred`,
-  );
-
-  const activeLabel = snapshot.activeSprintNumber !== null ? `Sprint ${snapshot.activeSprintNumber}` : "none";
-  lines.push(`**Sprints:** ${snapshot.sprintTotal} planned · active: ${activeLabel}`);
-  lines.push("");
-
-  if (snapshot.activeSprintNumber === null) {
-    lines.push("No active sprint.");
-  } else {
-    const totalItems = snapshot.activeSprintItems.length;
-    const doneItems = snapshot.activeSprintItems.filter((i) => i.status === "done").length;
-
-    lines.push(`### Sprint ${snapshot.activeSprintNumber} — ${snapshot.activeSprintGoal ?? ""}`);
-    lines.push(`Progress: ${snapshot.activeSprintPercentDone}% (${doneItems}/${totalItems} items done)`);
-    lines.push("");
-
-    for (const item of snapshot.activeSprintItems) {
-      const icon = statusIcon(item.status);
-      lines.push(`- ${icon} ${item.title} — ${item.status} · ${item.criteriaMet}/${item.criteriaTotal}`);
-    }
-  }
-
-  lines.push("");
-  lines.push("### Blockers");
-  if (snapshot.blockers.length === 0) {
-    lines.push("_None_");
-  } else {
-    for (const b of snapshot.blockers) {
-      lines.push(`- ${b.title}: ${b.reason}`);
-    }
-  }
-
-  lines.push("");
-  lines.push("### Worker");
-  lines.push(`Stage: ${snapshot.workerCurrentStage ?? "(idle)"}`);
-  lines.push(`Last event: ${snapshot.workerLastEventUtc ?? "(none)"}`);
-
-  return lines.join("\n");
-}

package/src/product-loop/reality-anchor.ts

@@ -1,45 +0,0 @@
-import type { Criterion } from "./types.js";
-
-export type { Criterion } from "./types.js";
-
-/**
- * Reality Anchor implements evidence verification for Definition-of-Done criteria.
- * It ensures that claims of "met" or "partial" status are backed by concrete evidence.
- */
-
-/**
- * Validates whether a given piece of evidence text matches one of the 5 allowed forms.
- * 1. file:line (e.g., src/sync.ts:42)
- * 2. test name (e.g., test('handles empty input'))
- * 3. commit sha (7-40 hex chars)
- * 4. benchmark (e.g., p95: 240ms)
- * 5. HTTP test (e.g., GET /api/users → 200)
- */
-export function evidenceLooksValid(text: string): boolean {
-  if (!text || text.trim().length === 0) return false;
-
-  const patterns = [
-    /\b\w+\.(ts|tsx|js|py|go|rs|java):\d+/,
-    /\btest\(['"`].+['"`]\)|describe\(['"`].+['"`]\)/,
-    /\b[a-f0-9]{7,40}\b/,
-    /\b(?:lighthouse|p95|p99|qps|throughput)[\s:=]+\d+/i,
-    /\b(GET|POST|PUT|DELETE|PATCH)\s+\/[^\s]+\s*→\s*\d{3}\b/,
-  ];
-
-  return patterns.some((regex) => regex.test(text));
-}
-
-/**
- * Annotates each criterion with evidenceValid: boolean based on the evidence field.
- */
-export function wrapSynthesisWithEvidence(criteria: Criterion[]): Criterion[] {
-  return criteria.map((c) => {
-    if (c.status === "unmet") {
-      return { ...c, evidenceValid: true }; // Unmet criteria don't need evidence
-    }
-    return {
-      ...c,
-      evidenceValid: c.evidence ? evidenceLooksValid(c.evidence) : false,
-    };
-  });
-}

package/src/product-loop/repo-audit.ts

@@ -1,314 +0,0 @@
-import { execSync } from "node:child_process";
-import { promises as fs } from "node:fs";
-import * as path from "node:path";
-
-/**
- * Deeper repo audit beyond {@link discoverProject}. The latter only detects
- * stack/test-runner from manifest files; this one walks the repo to give the
- * loop enough context to know whether the user is on a greenfield project, an
- * upgrade-existing-source job, or a scratch directory — and provides excerpts
- * the clarifier/debate can quote so its analysis is grounded in real code, not
- * generic boilerplate ("Review the current source code in src/").
- */
-export interface RepoAudit {
-  hasProject: boolean;
-  /**
-   *  - greenfield: nothing of substance in cwd
-   *  - upgrade-existing: src/ + tests + (README or docs) → established codebase
-   *  - scratch-dir: some files but no src+tests+docs combo
-   */
-  mode: "greenfield" | "upgrade-existing" | "scratch-dir";
-  topLevelDirs: string[];
-  srcFileCount: number;
-  testFileCount: number;
-  testFramework?: string;
-  hasCoverageConfig: boolean;
-  hasDocs: boolean;
-  readmeExcerpt?: string;
-  readmeSections?: string[];
-  packageMeta?: { name?: string; version?: string; description?: string };
-  recentCommits: string[];
-}
-
-const SRC_GLOB_DIRS = ["src", "lib", "app", "packages"];
-const TEST_DIR_HINTS = ["tests", "test", "__tests__", "spec"];
-const DOC_DIR_HINTS = ["docs", "doc", "documentation"];
-const COVERAGE_HINTS = [
-  "vitest.config.ts",
-  "vitest.config.js",
-  "jest.config.ts",
-  "jest.config.js",
-  ".coveragerc",
-  "pyproject.toml",
-];
-
-const CODE_EXTENSIONS = new Set([
-  ".ts",
-  ".tsx",
-  ".js",
-  ".jsx",
-  ".mjs",
-  ".cjs",
-  ".py",
-  ".go",
-  ".rs",
-  ".java",
-  ".kt",
-  ".rb",
-  ".php",
-]);
-const TEST_FILE_RE = /\.(test|spec)\.(ts|tsx|js|jsx|mjs|cjs|py)$/i;
-
-export async function auditRepo(cwd: string | undefined): Promise<RepoAudit> {
-  const audit: RepoAudit = {
-    hasProject: false,
-    mode: "greenfield",
-    topLevelDirs: [],
-    srcFileCount: 0,
-    testFileCount: 0,
-    hasCoverageConfig: false,
-    hasDocs: false,
-    recentCommits: [],
-  };
-  if (!cwd) return audit;
-
-  try {
-    const entries = await fs.readdir(cwd, { withFileTypes: true });
-    audit.topLevelDirs = entries.filter((e) => e.isDirectory() && !e.name.startsWith(".")).map((e) => e.name);
-  } catch {
-    return audit;
-  }
-
-  const lowerDirs = new Set(audit.topLevelDirs.map((d) => d.toLowerCase()));
-  audit.hasDocs = DOC_DIR_HINTS.some((d) => lowerDirs.has(d));
-
-  // Count src + test files (capped to avoid blowing up on huge repos)
-  for (const dir of SRC_GLOB_DIRS) {
-    if (!lowerDirs.has(dir)) continue;
-    const counts = await countCodeFiles(path.join(cwd, dir), 0, 5000);
-    audit.srcFileCount += counts.code;
-    audit.testFileCount += counts.tests;
-  }
-  for (const dir of TEST_DIR_HINTS) {
-    if (!lowerDirs.has(dir)) continue;
-    const counts = await countCodeFiles(path.join(cwd, dir), 0, 5000);
-    // Anything under tests/ counts as a test file regardless of naming.
-    audit.testFileCount += counts.code + counts.tests;
-  }
-
-  // Detect test framework and coverage config
-  for (const hint of COVERAGE_HINTS) {
-    const exists = await fileExists(path.join(cwd, hint));
-    if (!exists) continue;
-    audit.hasCoverageConfig = true;
-    if (/vitest/.test(hint)) audit.testFramework = "vitest";
-    else if (/jest/.test(hint)) audit.testFramework = "jest";
-    break;
-  }
-  const pkgRaw = await readIfExists(path.join(cwd, "package.json"));
-  if (pkgRaw) {
-    try {
-      const pkg = JSON.parse(pkgRaw);
-      audit.packageMeta = {
-        name: pkg.name,
-        version: pkg.version,
-        description: pkg.description,
-      };
-      if (!audit.testFramework) {
-        const all = { ...(pkg.dependencies ?? {}), ...(pkg.devDependencies ?? {}) };
-        if (all.vitest) audit.testFramework = "vitest";
-        else if (all.jest) audit.testFramework = "jest";
-        else if (all.mocha) audit.testFramework = "mocha";
-      }
-    } catch {
-      /* ignore */
-    }
-  }
-
-  // README
-  const readme = await findReadme(cwd);
-  if (readme) {
-    audit.readmeExcerpt = readme.excerpt;
-    audit.readmeSections = readme.sections;
-  }
-
-  // Recent commits — best-effort, skip silently when not a git repo or no git
-  try {
-    const out = execSync("git log --oneline -n 5 --no-decorate", {
-      cwd,
-      encoding: "utf-8",
-      stdio: ["ignore", "pipe", "ignore"],
-      timeout: 1500,
-    });
-    audit.recentCommits = out
-      .split("\n")
-      .map((l) => l.trim())
-      .filter(Boolean)
-      .slice(0, 5);
-  } catch {
-    /* ignore */
-  }
-
-  audit.hasProject = audit.srcFileCount > 0 || audit.testFileCount > 0 || !!audit.readmeExcerpt;
-
-  if (audit.srcFileCount >= 3 && audit.testFileCount >= 1 && (audit.hasDocs || !!audit.readmeExcerpt)) {
-    audit.mode = "upgrade-existing";
-  } else if (audit.hasProject) {
-    audit.mode = "scratch-dir";
-  } else {
-    audit.mode = "greenfield";
-  }
-
-  return audit;
-}
-
-async function countCodeFiles(dir: string, depth: number, budget: number): Promise<{ code: number; tests: number }> {
-  if (depth > 6 || budget <= 0) return { code: 0, tests: 0 };
-  let code = 0;
-  let tests = 0;
-  let entries: import("node:fs").Dirent[];
-  try {
-    entries = await fs.readdir(dir, { withFileTypes: true });
-  } catch {
-    return { code: 0, tests: 0 };
-  }
-  for (const e of entries) {
-    if (e.name.startsWith(".") || e.name === "node_modules" || e.name === "dist" || e.name === "build") continue;
-    const full = path.join(dir, e.name);
-    if (e.isDirectory()) {
-      const sub = await countCodeFiles(full, depth + 1, budget - code - tests);
-      code += sub.code;
-      tests += sub.tests;
-    } else if (e.isFile()) {
-      const ext = path.extname(e.name).toLowerCase();
-      if (!CODE_EXTENSIONS.has(ext)) continue;
-      if (TEST_FILE_RE.test(e.name)) tests++;
-      else code++;
-    }
-    if (code + tests >= budget) break;
-  }
-  return { code, tests };
-}
-
-async function findReadme(cwd: string): Promise<{ excerpt: string; sections: string[] } | null> {
-  const candidates = ["README.md", "Readme.md", "readme.md", "README", "README.txt"];
-  for (const c of candidates) {
-    const raw = await readIfExists(path.join(cwd, c));
-    if (!raw) continue;
-    const lines = raw.split("\n");
-    // Excerpt: first 2 non-empty paragraphs (capped 500 chars)
-    const paragraphs: string[] = [];
-    let buf: string[] = [];
-    for (const line of lines) {
-      if (!line.trim()) {
-        if (buf.length > 0) {
-          paragraphs.push(buf.join(" ").trim());
-          buf = [];
-          if (paragraphs.length >= 2) break;
-        }
-      } else if (!line.startsWith("#")) {
-        buf.push(line.trim());
-      }
-    }
-    if (buf.length > 0 && paragraphs.length < 2) paragraphs.push(buf.join(" ").trim());
-    const excerpt = paragraphs.join("\n\n").slice(0, 500);
-    const sections = lines
-      .filter((l) => /^#{1,3}\s+/.test(l))
-      .slice(0, 12)
-      .map((l) => l.replace(/^#{1,3}\s+/, "").trim());
-    return { excerpt, sections };
-  }
-  return null;
-}
-
-async function readIfExists(p: string): Promise<string | null> {
-  try {
-    return await fs.readFile(p, "utf-8");
-  } catch {
-    return null;
-  }
-}
-
-async function fileExists(p: string): Promise<boolean> {
-  try {
-    await fs.stat(p);
-    return true;
-  } catch {
-    return false;
-  }
-}
-
-/**
- * Produce extra prefills for SEED_DIMENSIONS based on the audit. These layer
- * on top of {@link discoverProject}'s prefills — caller merges them with first
- * write wins so manifest-derived stack/test info is not overwritten.
- */
-export function additionalPrefills(a: RepoAudit): Map<string, string> {
-  const out = new Map<string, string>();
-  if (!a.hasProject) return out;
-
-  // persona — when the codebase has README + tests it is overwhelmingly a
-  // developer-facing tool/library. Confidence is high enough to skip the Q.
-  if ((a.readmeExcerpt || a.readmeSections?.length) && a.testFileCount > 0) {
-    out.set("persona", "developers using this codebase");
-  }
-
-  // success-metric — refine when we know the test framework
-  if (a.testFramework && a.hasCoverageConfig) {
-    out.set("success-metric", `100% test pass with >80% coverage (runner: ${a.testFramework})`);
-  }
-
-  // core-features — for upgrade-existing mode the user's idea is "improve
-  // existing src", so MVP = "match existing project surface area".
-  if (a.mode === "upgrade-existing") {
-    out.set("core-features", "match existing project surface area (upgrade in place)");
-  }
-
-  return out;
-}
-
-/**
- * Multi-line context block that the clarifier and debate stages should
- * receive as `conversationContext`. Grounds prompts in real repo state so
- * Clarified Spec / Research output reference actual files instead of
- * generic boilerplate.
- */
-export function auditAsContextBlock(a: RepoAudit): string {
-  if (!a.hasProject) return "Cwd appears to be a greenfield/empty directory.";
-  const lines: string[] = ["## Repository audit"];
-  lines.push(`Mode: ${a.mode}`);
-  if (a.packageMeta?.name) {
-    lines.push(`Package: ${a.packageMeta.name}${a.packageMeta.version ? `@${a.packageMeta.version}` : ""}`);
-    if (a.packageMeta.description) lines.push(`Description: ${a.packageMeta.description}`);
-  }
-  lines.push(`Top-level dirs: ${a.topLevelDirs.join(", ")}`);
-  lines.push(`Source files: ${a.srcFileCount}, test files: ${a.testFileCount}`);
-  if (a.testFramework)
-    lines.push(`Test runner: ${a.testFramework}${a.hasCoverageConfig ? " (coverage configured)" : ""}`);
-  if (a.readmeExcerpt) {
-    lines.push("");
-    lines.push("README excerpt:");
-    lines.push(a.readmeExcerpt);
-  }
-  if (a.readmeSections?.length) {
-    lines.push("");
-    lines.push(`README sections: ${a.readmeSections.slice(0, 8).join(" / ")}`);
-  }
-  if (a.recentCommits.length) {
-    lines.push("");
-    lines.push("Recent commits:");
-    for (const c of a.recentCommits) lines.push(`- ${c}`);
-  }
-  return lines.join("\n");
-}
-
-export function formatAuditSummary(a: RepoAudit): string | null {
-  if (!a.hasProject) return null;
-  const bits: string[] = [`mode=**${a.mode}**`];
-  if (a.packageMeta?.name) bits.push(`pkg \`${a.packageMeta.name}\``);
-  bits.push(`src=${a.srcFileCount}`);
-  bits.push(`tests=${a.testFileCount}`);
-  if (a.testFramework) bits.push(`runner=${a.testFramework}`);
-  if (a.hasDocs) bits.push("docs ✓");
-  return `**Repo audit:** ${bits.join(" · ")}`;
-}