Weekly Review

Generates a weekly review without asking questions. Scans vault data and writes a [!weekly-review]- block into the weekly note, then appends a minimal space for the user's own reflections.

Language Detection

Before generating any content, detect the user's writing language. Read the 5 most recent daily notes (periodic/daily/YYYY-MM-DD.md). In non-frontmatter, non-wikilink, non-tag text, count Cyrillic vs Latin characters. If Cyrillic > 30% → write the entire review in Russian. Otherwise → English.

Apply to ALL section headers, metric labels, inference lines (→), and participation prompts. Callout types ([!weekly-review]) and wikilinks remain unchanged regardless of detected language.

Weekly Note Format

Path: periodic/weekly/YYYY-WWW.md

Compute week and parent month (cross-platform, using obsidian eval):

# Current ISO 8601 week e.g. 2026-W08
obsidian eval code="(function(){const d=new Date();const dt=new Date(Date.UTC(d.getFullYear(),d.getMonth(),d.getDate()));const day=dt.getUTCDay()||7;dt.setUTCDate(dt.getUTCDate()+4-day);const y=dt.getUTCFullYear();const ys=new Date(Date.UTC(y,0,1));const w=Math.ceil(((dt-ys)/86400000+1)/7);return y+'-W'+String(w).padStart(2,'0')})()"

# Parent month e.g. 2026-02
obsidian eval code="const d=new Date();d.getFullYear()+'-'+String(d.getMonth()+1).padStart(2,'0')"

Frontmatter (exact from template):

---
tags:
  - periodic/week
up:
  - "[[periodic/monthly/YYYY-MM|YYYY-MM]]"
created: YYYY-MM-DDTHH:mm:ssZ
updated: YYYY-MM-DDTHH:mm:ssZ
reviewed: false
cssclasses:
  - hide-backlinks
---

> [!success]- 🔻 history 🔻
> `$=await dv.view("templates/views/periodic", {type: "day"})`

If the weekly note already exists — find the [!weekly-review]- block and replace it. Otherwise create the note with the template above, then append the review block after [!success]-.

Execution Flow

User: "weekly review" / "итоги недели" / "недельный обзор"
    |
1. Compute current ISO week (YYYY-WWW) and date range Mon–Sun
    |
2. Find or create periodic/weekly/YYYY-WWW.md
    |
3. Detect language from last 5 daily notes
    |
4. Read 7 daily notes for Mon–Sun (skip missing days silently)
   Also extract: free-text user writing in each note (content OUTSIDE [!briefing]- callout,
   not task lines, not pure wikilinks — what the user actually wrote themselves)
    |
5. Gather IN PARALLEL:
   a. Completed tasks — CLI: obsidian tasks done / obsidian search query="✅ YYYY-MM"
   b. Themes — count wikilinks across daily notes
   c. Project progress — read active projects via obsidian search + mentions in daily notes
   d. Open threads — extract Threads sections from briefing callouts
   e. Journal patterns — aggregate free-text themes from user writing in daily notes
    |
6. For each section: collect data → derive one "→" inference from the specific pattern
    |
7. Generate review callout with inferences
    |
8. Write [!weekly-review]- callout into weekly note
    |
9. Append user participation section below the callout (outside it)
    |
10. Minimal chat confirmation

Completed Tasks

Search entire vault for tasks completed this week:

obsidian tasks done format=json                    # all done tasks with metadata
obsidian search query="✅ YYYY-MM" path=periodic  # count by month (scoped to periodic notes)

Filter: keep only lines where ✅ date falls within Mon–Sun of current week

Extract human-readable task text: strip - [x], #task/*, #category/*, #priority/*, and all emoji date markers (📅, ⏰, ✅ YYYY-MM-DD). Keep only the description.

List as plain bullets with abbreviated weekday (Mon/Tue/Wed/Thu/Fri/Sat/Sun in English, пн/вт/ср/чт/пт/сб/вс in Russian):

> **Completed tasks (5):**
> - Write chapter outline for the longform project — Mon
> - Set up automated backup script — Wed
> - Publish draft article — Fri

Never duplicate task markdown — no - [x], no tags, no emoji. Only the human-readable description + day.

→ Inference: After listing tasks, derive one data-specific observation. Look for:

Tasks clustered on 1–2 days → what blocked the other days?
One task category dominates (all admin, all writing) → imbalance or intentional sprint?
Fewer tasks than usual → rest week or something blocked execution?
Many tasks completed but no project progress → lots of action, little forward movement

Write the inference in detected language. Skip if no clear pattern.

If nothing found — skip section silently.

Hot Themes

Parse all wikilinks from the 7 daily notes. Count frequency of links pointing to:

base/_meta-notes/
base/notes/

Show top 3–5 by count. Skip structural navigation links (monthly, weekly, project index pages).

> **Hot themes:**
> - [[knowledge management]] — 6 mentions
> - [[writing process]] — 4 mentions
> - [[productivity systems]] — 2 mentions

→ Inference: After listing themes, derive one data-specific observation. Look for:

Top theme mentioned 4+ times but no new note created this week → thinking without writing
Many themes at low count (all 2–3) → scattered attention, no single focus this week
One theme dominates (>50% of all mentions) → deep dive or tunnel vision?
Theme appears only in the last 2 days → momentum building that could carry into next week

Write the inference in detected language. Skip if no clear pattern.

Journal Patterns

Read free-text content in each daily note — text that is NOT inside the [!briefing]- callout, NOT task lines (- [ ] / - [x]), and NOT pure wikilink lists. This is what the user wrote themselves: observations, thoughts, diary entries.

If a concept, concern, or question recurs in user-written text across 3+ different days — surface it as a pattern:

> **Journal pattern:**
> You kept returning to [topic/concept] across N days — no note yet, but the thinking is accumulating.

Skip this section if:

The user wrote very little free text this week
No pattern across multiple days is visible
The pattern duplicates what's already captured in Hot Themes

Do not quote raw journal text. Only summarize the pattern.

Project Progress

obsidian search query="tag:#project [status: 🟦]" and obsidian search query="tag:#project [status: 🟥]" → list active projects
For each project, search its title in the 7 daily notes via obsidian search:context query="[[Project Name]]" path=periodic/daily
Extract: decisions made, blockers, progress, next steps from those mentions
Write as narrative: status emoji + 1 sentence of what happened this week

> **Projects:**
> - [[Personal Blog Relaunch]] 🟦 — drafted two new posts, design still blocked
> - [[Note-taking Workflow v2]] 🟦 — finalized the capture stage, review stage pending
> - [[Podcast Pilot Episode]] 🟥 — no progress this week

If a project had no mentions this week — include with status only, no narrative.

→ Inference: After listing projects, derive one data-specific observation. Look for:

Same blocker mentioned 2+ weeks running → the blocker is the real issue, not the project
Only one project received attention → others are being starved
Project had no progress but was active the previous week → what interrupted it?

Write the inference in detected language. Skip if no clear pattern.

Open Threads

Extract **Threads:** section from [!briefing]- callouts in each of the 7 daily notes.

A thread is "open" if:

Appears in Threads sections across 2+ different daily notes
No matching ✅ completion found in vault for that thread

Sort by persistence (most days appearing first):

> **Open threads:**
> - Respond to collaboration request — 5 days
> - Schedule dentist appointment — 3 days
> - Draft reply to forum post — 2 days

→ Inference: After listing threads, derive one data-specific observation. Look for:

Thread persisting 5+ days → is it actually a priority or can it be closed?
Multiple threads at exactly 2 days → fragmented attention; many small things unresolved
All threads are low-friction actions (scheduling, replying) → tendency to defer easy tasks

Write the inference in detected language. Skip if no clear pattern.

Reflection Questions

Generate exactly 2 sharp questions grounded in the week's data. Use the accumulated "→" inferences as the primary source — the two most significant patterns of the week become the two questions.

Additional patterns to draw from:

Pattern	Question type
Theme with high mentions but no new note	"What's blocking the conversion into a note?"
Project same status 2+ weeks	"What specifically is needed for the next step?"
Thread persisting 4+ days without resolution	"Is this a real priority or can it be closed?"
Tasks clustered on 1–2 days	"What was preventing work on the other days?"
One project dominates all mentions	"Is this intentional or are other projects getting lost?"
User journal mentions concept but no wikilink exists	"Is this concept ready to become a note?"

> **Reflection:**
> 1. [[writing process]] appeared 6 times this week but no new note was created — what's blocking the conversion from thinking to writing?
> 2. [[Personal Blog Relaunch]] has been "design blocked" for two weeks — what's the actual next action to unblock it?

Questions must be specific: name the actual note, project, or number from this week's data. Never generic.

Output Format

Full callout written into the weekly note:

> [!weekly-review]- Weekly review — YYYY-WWW
> **Completed tasks (N):**
> - Task description — weekday
> → [inference about task pattern — only if pattern found]
>
> **Hot themes:**
> - [[note-or-meta]] — N mentions
> → [inference about theme pattern — only if pattern found]
>
> **Journal pattern:** [one sentence — only if cross-day pattern found]
>
> **Projects:**
> - [[Project Name]] 🟦 — narrative from the week
> → [inference about project portfolio — only if pattern found]
>
> **Open threads:**
> - Thread description — N days
> → [inference about thread accumulation — only if pattern found]
>
> **Reflection:**
> 1. Specific question grounded in data
> 2. Specific question grounded in data

Rules:

Write into note — never just output to chat
Collapsed by default (- suffix on callout)
Callout title includes the week ID: [!weekly-review]- Weekly review — 2026-W08
All labels translated to detected language (e.g. "Задачи завершены", "Горячие темы", etc.)
Skip sections with no content — don't write empty headers
Skip "→" inference lines where no clear pattern exists — never write generic "→" lines
Every reference = [[wikilink]], never vague "in one of the notes"

User Participation Sections

After the [!weekly-review]- callout, appended outside it (in the note body), with a horizontal rule separator:

Default form (always):

---
**Мои выводы / My conclusions:**

**Фокус на следующую неделю / Focus for next week:**

Extended form (when review data shows a specific sharp pattern): Replace the section above with:

---
*[One direct question from the review's most significant finding — e.g., "Что блокировало [[writing process]] от превращения в заметку?" / "What's blocking [[writing process]] from becoming a note?"]*

**Мои выводы / My conclusions:**

**Фокус на следующую неделю / Focus for next week:**

Write the italic question in detected language. Use extended form only when a pattern is unambiguous. Default to the plain two-header form otherwise.

Note: The "Focus for next week" section is read by the monthly-review skill when computing Planned vs Actual.

Chat Output

After writing, confirm minimally:

Review written → periodic/weekly/2026-W08.md

Tasks: 5 completed.
Themes: knowledge management (6), writing process (4).
Projects: 3 active.
Threads: 3 open.

If the note was created — first line: Created periodic/weekly/2026-W08.md + review. Adapt language of confirmation to match detected language.