Weekly ReviewInformation / Field Journal
A trusted system, revisited
Make room
for next week.
Find commitments hiding in your notes. Name what is missing. Leave with one concrete next action.
01 / InputsGet clearCollect, clarify, route.
02 / CommitmentsGet currentLists, calendars, projects.
03 / PossibilitiesGet creativeReview ideas. Choose next.
Bring
Notes and commitments.
A dedicated notes folder, task lists, waiting-for items, and your own previous/upcoming calendar context.
Use only records you are authorized to review. No live calendar connection is required or included.
Leave with
An honest review record.
An open-loop inventory, a ten-step checklist verdict with named gaps, and a transparent commitment-health audit.
Scripts handle inventory and checks. You and the assistant handle judgment.
Personal maintenance
For a full week.
Or a fresh restart.
- Freelancers balancing client promises and personal projects.
- Operators whose next actions are scattered across notes and lists.
- Anyone restarting a lapsed weekly review without a back-processing marathon.
A personal commitment review, not a team sprint retrospective, automatic reminder service, or wellbeing measure.
The included toolkit
Three small scripts.
No extra service to connect.
The package includes the original Skill, three Python scripts, three references, and two templates/examples. Scripts use only Python 3 standard library and make no model or network calls.
No additional API key is required. An AI assistant with file access and Python execution still guides the review; model access or subscription charges may apply. Open SKILL.md in your file-capable assistant or use its supported Skill installation flow.
Actual local validation
3original scripts executed
12normal and boundary cases
0unexpected case failures
Synthetic notes produced three unchecked boxes, two TODO/FIXME markers, and one stale file. Four commitments produced a 57.5/100 DRIFTING heuristic verdict.
Counts can overlap. They are not unique-task totals. The score is not objective productivity measurement or proof that actual review steps occurred.
Keep the verdict honest
The gate checks declared done/skip inputs. A reasoned skip can yield COMPLETE, including a core step, and emits a warning. Use normal invocation rather than --sample, whose illustrative gate mode always exits 0.
The scanner can follow file symlinks: use a dedicated notes folder without them. No automatic email/calendar sync, outgoing message, or background reminder is included. Only Codex/Python was validated; Claude and Claude.ai were not tested.
Source: alirezarezvani/claude-skills, commit 19392f7a08264ed00486a251f5b2098321771f94. MIT, Copyright (c) 2025 Alireza Rezvani. Nine original files and method attributions are preserved unchanged.