weekly-review · diff
git:20261007.e522be7 to git:20261014.67a9e0a
53 added, 41 removed. Audit A to A.
---
name: weekly-review
- description: Retrospective review of an existing completed weekly Obsidian note.
+ description: Summarize a completed week's daily notes into key findings, potential atomic notes, and outstanding action items.
disable-model-invocation: true
argument-hint: '[ISO week — omit for the most recently completed week]'
---
# Weekly Review
- Review what happened in one completed week. This skill does not create the next weekly plan or edit
- monthly, quarterly, or yearly goals.
+ Summarize one calendar week's daily notes and write the result into that week's `## Weekly Review`
+ section. This is a retrospective, not a next-week planning workflow.
## 1. Confirm the review period
- Read `AGENTS.md`, resolve the requested or most recently completed ISO week, and identify its weekly
- note from the live vault convention. Ask when a partial or current week makes the period ambiguous.
+ Read the vault's `AGENTS.md` and `../vault/SKILL.md` for live paths, ownership, and safe-write rules.
+ Resolve the requested ISO week or default to the most recently completed Monday–Sunday week. Locate
+ its existing weekly note from the live vault convention and compute the exact seven-day range.
+ Resolve each daily note by its own calendar date; an ISO week can cross a calendar-year folder
+ boundary. Archived journals may be read as evidence, but archived weekly notes remain read-only.
- The weekly note must already exist and be complete enough to review. If it is missing or still being
- written, stop and ask the user to finish it first.
+ If the weekly note is missing, stop and report the gap. Do not create or reconstruct it as a side
+ effect. Ask when the requested period is partial or ambiguous. Confirm that the live vault contract
+ permits `weekly-review` to manage only `## Weekly Review`; otherwise return the draft without writing.
- Completion: one existing completed weekly note and exact date range are confirmed.
+ Completion: one existing weekly note, its exact Monday–Sunday range, and the write boundary for
+ `## Weekly Review` are confirmed.
- ## 2. Gather evidence
+ ## 2. Gather bounded evidence
- Read:
+ Read the weekly note and each existing `YYYY-MM-DD.md` daily note in the confirmed date range, in
+ chronological order. Follow links from those notes only when needed to understand an entry.
- - the weekly note;
- - daily notes within its date range;
- - notes explicitly linked from those notes;
- - active goals and projects only when needed to evaluate stated alignment.
+ Extract:
- Extract original commitments, outcomes, unfinished work, decisions, blockers, friction, unplanned
- wins, and lessons. Cite a source note for each claim. Missing notes are evidence gaps, not proof of
- inactivity.
+ | Category | Evidence to find |
+ | --- | --- |
+ | Key findings | Insights, learnings, realizations, decisions, and useful links |
+ | Potential atomic notes | Recurring themes, surprising insights, new connections, or durable ideas |
+ | Action items | Outstanding tasks, TODOs, and follow-ups |
- Completion: each proposed review claim has a source or an explicit user correction.
+ A potential atomic note should be worth remembering beyond the reviewed week. Deduplicate action
+ items and exclude work already completed. Missing or sparse notes are evidence gaps, not proof that
+ nothing happened. Use external systems only when the user explicitly requests them.
- ## 3. Close each commitment
+ Completion: every proposed item traces to a weekly or daily note, a linked source, or an explicit user
+ correction.
- For every commitment recorded in the weekly note, have the user confirm one state:
+ ## 3. Draft Weekly Review
- - complete;
- - intentionally carried forward, with its destination;
- - retired.
+ Read the active Weekly template to confirm the review headings, without changing it. If its structure
+ differs from the one below, show the discrepancy and agree on the target before writing.
- Keep unplanned accomplishments under Unplanned Wins rather than rewriting the original plan.
+ Draft this review structure:
- Completion: every original commitment has one confirmed closure state.
+ ```markdown
+ ## Weekly Review
- ## 4. Draft the retrospective
+ ### Key Findings
+ - [Finding with relevant [[wikilinks]] inline]
- Prepare only:
+ ### Potential Atomic Notes
+ - **[Proposed title as insight]**: [why it deserves its own note]
- - commitment outcomes with evidence;
- - unplanned wins;
- - decisions and lessons worth preserving;
- - friction;
- - Keep / Change / Remove;
- - confirmed carry-forward candidates.
+ ### Action Items
+ - [ ] [Outstanding follow-up]
+ ```
- Report goal alignment as evidence, but do not revise goals. When evidence suggests an objective
- needs changing, offer `/goals` after the review.
+ Keep all three subsections. Replace example bullets with the week's evidence; use `- None identified.`
+ when a subsection has no supported items. Keep relevant links inline instead of adding a separate link
+ dump. Potential atomic notes are proposals only: do not create them. Action items are outstanding
+ follow-ups, not automatic commitments for the next week.
- ## 5. Approve, write, and verify
+ Completion: the draft is concise, deduplicated, and every factual statement is supported.
- Show the complete retrospective draft before writing. After approval, update only the reviewed
- weekly note through the `vault` safe-write flow. Preserve the original plan and existing
- human-written content. Reread and verify that each commitment has exactly one closure state.
+ ## 4. Approve, write, and verify
- Return the closure summary, unplanned wins, Keep / Change / Remove, confirmed carry-forward, and
- unresolved evidence gaps.
+ Show the exact proposed `## Weekly Review` section and obtain explicit approval through the `vault`
+ safe-write flow. If the section already exists, replace only its three review subsections. If it is
+ absent, append the complete section. Preserve `Focus for this week` and every other existing section.
+
+ Reread the affected weekly note. Verify that there is exactly one `## Weekly Review` section, its
+ three subsections match the approved draft, and no other content changed. Return the written review,
+ evidence gaps, and whether the note changed.