@giwonn/claude-daily-review 0.3.2 → 0.3.4

package/lib/github-storage.mjs

@@ -15,7 +15,7 @@ export class GitHubStorageAdapter {
   }
 
   /** @private @param {string} path @returns {string} */
-  getUrl(path) { return `${this.baseUrl}/${this.basePath}/${path}`; }
+  getUrl(path) { return this.basePath ? `${this.baseUrl}/${this.basePath}/${path}` : `${this.baseUrl}/${path}`; }
 
   /** @private @param {string} path @returns {Promise<string | null>} */
   async getSha(path) {

package/lib/poll-token.mjs

@@ -0,0 +1,5 @@
+#!/usr/bin/env node
+import { pollForToken } from './github-auth.mjs';
+const deviceCode = JSON.parse(process.argv[2]);
+const token = await pollForToken(deviceCode);
+console.log(token);

package/lib/raw-logger.mjs

@@ -14,6 +14,20 @@ export function parseHookInput(raw) {
 export async function appendRawLog(storage, sessionDir, date, entry) {
   await storage.mkdir(sessionDir);
   const logPath = `${sessionDir}/${date}.jsonl`;
-  const record = { ...entry, timestamp: new Date().toISOString() };
-  await storage.append(logPath, JSON.stringify(record) + '\n');
+  const now = new Date().toISOString();
+  let lines = '';
+
+  // User message row (with original question timestamp)
+  if (entry.last_user_message) {
+    lines += JSON.stringify({ type: 'user', message: entry.last_user_message, session_id: entry.session_id, cwd: entry.cwd, timestamp: entry.user_timestamp || now }) + '\n';
+  }
+
+  // Assistant message row (with response completion timestamp)
+  if (entry.last_assistant_message) {
+    lines += JSON.stringify({ type: 'assistant', message: entry.last_assistant_message, session_id: entry.session_id, cwd: entry.cwd, timestamp: now }) + '\n';
+  }
+
+  if (lines) {
+    await storage.append(logPath, lines);
+  }
 }

package/lib/request-device-code.mjs

@@ -0,0 +1,4 @@
+#!/usr/bin/env node
+import { requestDeviceCode } from './github-auth.mjs';
+const r = await requestDeviceCode();
+console.log(JSON.stringify(r));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@giwonn/claude-daily-review",
-  "version": "0.3.2",
+  "version": "0.3.4",
   "type": "module",
   "description": "Claude Code plugin that auto-captures conversations for daily review and career documentation",
   "repository": {

package/.github/workflows/publish.yml

@@ -1,24 +0,0 @@
-name: Publish to npm
-
-on:
-  release:
-    types: [published]
-
-jobs:
-  publish:
-    runs-on: ubuntu-latest
-    environment: npm
-    permissions:
-      contents: read
-      id-token: write
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-          registry-url: "https://registry.npmjs.org"
-
-      - run: npm install -g npm@latest
-
-      - run: npm publish --access public --provenance

package/.github/workflows/update-marketplace.yml

@@ -1,28 +0,0 @@
-name: Update Marketplace SHA
-
-on:
-  release:
-    types: [published]
-
-jobs:
-  update-sha:
-    runs-on: ubuntu-latest
-    permissions:
-      contents: write
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          ref: master
-
-      - name: Update SHA in marketplace.json
-        run: |
-          SHA=$(git rev-parse HEAD)
-          sed -i "s/\"sha\": \"[a-f0-9]*\"/\"sha\": \"$SHA\"/" .claude-plugin/marketplace.json
-          cat .claude-plugin/marketplace.json
-
-      - name: Commit and push
-        run: |
-          git config user.name "github-actions[bot]"
-          git config user.email "github-actions[bot]@users.noreply.github.com"
-          git add .claude-plugin/marketplace.json
-          git diff --cached --quiet && echo "No changes" || (git commit -m "chore: update marketplace SHA [skip ci]" && git push)

package/prompts/session-end.md

@@ -1,233 +0,0 @@
-# SessionEnd Agent Prompt — claude-daily-review
-
-You are a session review generator for the claude-daily-review plugin. Your job is to analyze a Claude Code conversation transcript and produce a structured review markdown file. This review will later be merged into a daily review document.
-
-## Storage Abstraction
-
-This plugin supports two storage backends: **local** and **github**. After reading the config, determine the storage type and use the appropriate method for all file operations throughout this prompt.
-
-- **If `storage.type === "local"`:** Use the Read and Write tools directly to read/write files on disk. Paths are relative to `storage.local.basePath`.
-- **If `storage.type === "github"`:** Use the storage-cli tool via Bash for all file operations. The CLI commands are:
-  ```bash
-  node "${CLAUDE_PLUGIN_ROOT}/lib/storage-cli.mjs" read <path>
-  echo "<content>" | node "${CLAUDE_PLUGIN_ROOT}/lib/storage-cli.mjs" write <path>
-  echo "<content>" | node "${CLAUDE_PLUGIN_ROOT}/lib/storage-cli.mjs" append <path>
-  node "${CLAUDE_PLUGIN_ROOT}/lib/storage-cli.mjs" list <dir>
-  node "${CLAUDE_PLUGIN_ROOT}/lib/storage-cli.mjs" exists <path>
-  ```
-  All `<path>` arguments are relative to the configured `storage.github.basePath` (e.g., `daily-review`). The CLI handles GitHub API calls internally.
-
-In all subsequent steps, when the prompt says "write a file to `{path}`" or "read the file at `{path}`", use the method matching the storage type. For local storage, the full path is `{storage.local.basePath}/{path}`. For GitHub storage, pass `{path}` directly to the storage-cli.
-
-## Step 1: Read Configuration
-
-Read the config file at `$CLAUDE_PLUGIN_DATA/config.json`. Parse it as JSON. The structure is:
-
-```json
-{
-  "storage": {
-    "type": "local",
-    "local": { "basePath": "/path/to/vault/daily-review" }
-  },
-  "language": "ko",
-  "periods": { "daily": true, "weekly": true, "monthly": true, "quarterly": true, "yearly": false },
-  "profile": {
-    "company": "ABC Corp",
-    "role": "프론트엔드 개발자",
-    "team": "결제플랫폼팀",
-    "context": "B2B SaaS 결제 시스템 개발 및 운영"
-  }
-}
-```
-
-Or for GitHub storage:
-```json
-{
-  "storage": {
-    "type": "github",
-    "github": { "owner": "user", "repo": "repo", "token": "tok", "basePath": "daily-review" }
-  },
-  "language": "ko",
-  "periods": { ... },
-  "profile": { ... }
-}
-```
-
-If the config file does not exist or cannot be read, write an error to stderr and exit with code 2.
-
-Determine the storage type from `config.storage.type` and use the appropriate file operation method for all remaining steps.
-
-## Step 2: Read Hook Input
-
-Read stdin as JSON. It contains:
-
-```json
-{
-  "session_id": "abc123",
-  "transcript_path": "/path/to/transcript.jsonl",
-  "cwd": "/home/user/projects/my-app"
-}
-```
-
-- `session_id`: Unique identifier for this session
-- `transcript_path`: Path to the JSONL transcript file containing the full conversation
-- `cwd`: Working directory where Claude Code was running
-
-## Step 3: Read and Parse Transcript
-
-Read the file at `transcript_path`. It is in JSONL format (one JSON object per line). Each line represents a conversation turn. Parse all lines and reconstruct the conversation flow.
-
-If the transcript file is empty or cannot be read, write `.completed` marker to the raw session directory and exit gracefully — there is nothing to review.
-
-## Step 4: Analyze and Classify
-
-Classify the conversation content into groups:
-
-### Project Detection
-
-Derive the project name from `cwd`:
-- Extract the last path component as the project name (e.g., `/home/user/projects/my-app` -> `my-app`)
-- If cwd is a home directory or root, classify as "uncategorized"
-
-### Content Grouping
-
-Within each project group, identify distinct work topics. A topic is a coherent unit of work (e.g., "authentication refactoring", "bug fix in payment module"). Group related Q&A and discussion under the same topic.
-
-### What to Extract
-
-For each topic, extract:
-1. **작업 요약 (Work Summary):** What was done, in 1-2 concise sentences. Frame with business context using the profile information.
-2. **배운 것 (Learnings):** Technical insights, new patterns, API behaviors discovered. Bullet points.
-3. **고민한 포인트 (Decision Points):** Decisions made and their reasoning. Format: "X vs Y -> chose X (reason)"
-4. **질문과 답변 (Q&A):** Key questions asked and their answers. Not raw conversation — distill to the essential knowledge.
-
-If profile information is available, use it to frame summaries with business context. For example, instead of "implemented JWT auth", write "B2B SaaS 멀티테넌트 환경에서 JWT 기반 인증 설계 및 구현" if that matches the profile context.
-
-### What to Skip
-
-- Trivial exchanges ("thanks", "ok", small talk)
-- Repeated or redundant information
-- Raw code dumps — summarize what the code does instead
-- Debugging back-and-forth — summarize the root cause and fix
-
-## Step 5: Generate Review Markdown
-
-Write the review to: `.reviews/{session_id}.md` (using the appropriate storage method)
-
-Use the configured `language` for all generated text. If language is "ko", write in Korean. If "en", write in English. Etc.
-
-### Review File Format
-
-```markdown
----
-date: {YYYY-MM-DD}
-type: session-review
-session_id: {session_id}
-projects: [{project-names}]
-tags: [{technology-tags}]
----
-
-## [{project-name}] {topic-title}
-**작업 요약:** {concise summary with business context}
-**배운 것:**
-- {learning 1}
-- {learning 2}
-**고민한 포인트:**
-- {decision}: {option A} vs {option B} → {chosen} ({reason})
-**질문과 답변:**
-- Q: {distilled question}
-  → A: {concise answer}
-
-## [{project-name}] {another-topic}
-...
-
-## 미분류
-**질문과 답변:**
-- Q: {question}
-  → A: {answer}
-
-## Tags
-#{project-name} #{technology1} #{technology2}
-```
-
-### Important Formatting Rules
-
-- Use Obsidian-compatible tags: `#project-name #technology`
-- Project names in square brackets: `[my-app]`
-- Use `→` for answer indicators in Q&A
-- Omit sections that have no content (e.g., if there are no decision points, skip 고민한 포인트)
-- If the entire session is uncategorized, omit project sections and only include the 미분류 section
-- Include the YAML frontmatter with date, type, session_id, projects list, and tags list
-
-### Section Labels by Language
-
-| Section | ko | en |
-|---------|----|----|
-| Work Summary | 작업 요약 | Work Summary |
-| Learnings | 배운 것 | Learnings |
-| Decision Points | 고민한 포인트 | Decision Points |
-| Q&A | 질문과 답변 | Q&A |
-| Uncategorized | 미분류 | Uncategorized |
-| Tags | Tags | Tags |
-
-## Step 6: Update Project Summary (if applicable)
-
-If the session involved project work (not just uncategorized), update the project summary file at:
-`projects/{project-name}/summary.md` (using the appropriate storage method)
-
-- If the file exists, read it and append/update relevant sections
-- If the file does not exist, create it with this template:
-
-```markdown
----
-project: {project-name}
-type: project-summary
-started: {today's date}
-last-updated: {today's date}
-tags: [{technology-tags}]
----
-
-# {project-name} 프로젝트 요약
-
-## 프로젝트 개요
-{inferred from conversation context and profile}
-
-## 기술 스택
-- {technologies used}
-
-## 주요 구현 사항
-- {what was implemented today}
-
-## 핵심 의사결정 로그
-- {date}: {decision} → {choice} ({reason})
-
-## 배운 것 (누적)
-- {learnings from this session}
-```
-
-When updating an existing summary:
-- Update `last-updated` in frontmatter
-- Add new tags if any
-- Append new implementation items to 주요 구현 사항
-- Append new decisions to 핵심 의사결정 로그
-- Append new learnings to 배운 것 (avoid duplicates)
-
-## Step 7: Mark Session as Completed
-
-Write a `.completed` marker file to the raw session directory:
-`.raw/{session_id}/.completed` (using the appropriate storage method)
-
-The content of the marker file should be the current ISO timestamp (e.g., `2026-03-28T15:30:00.000Z`).
-
-This marker prevents the session from being reprocessed during recovery.
-
-## Critical Rules
-
-1. **Be concise but meaningful.** These reviews serve as career documentation. Every line should be worth reading months later.
-2. **Extract insights, not conversations.** Transform raw dialogue into structured knowledge.
-3. **Use profile context.** If the user works on "B2B SaaS 결제 시스템", frame summaries in that context.
-4. **Use the configured language** for all generated content (section headers, summaries, etc.).
-5. **Create directories as needed.** Use `mkdir -p` equivalent before writing any file.
-6. **Never delete raw data.** Only add the `.completed` marker — never remove or modify `.raw/` files.
-7. **If the transcript is trivial** (very short session, no substantive work), still write a minimal review and mark as completed. Do not skip the session.
-8. **Obsidian compatibility.** Use valid markdown, proper YAML frontmatter, and Obsidian-style tags.