The MCP Planning Loop: A Framework for Using Beyond Time with Claude

An MCP connection to Beyond Time gives Claude live access to your goal data. That’s the infrastructure. But infrastructure without a rhythm produces noise, not results.

We call the rhythm the MCP Planning Loop — a three-cadence structure that turns the Beyond Time connection into a genuine planning practice. Each cadence uses a different subset of MCP commands and serves a different purpose. Together they close the gap between intention (setting goals) and execution (tracking and adjusting them).

Why a Framework Is Necessary

It’s tempting to assume that a live data connection makes planning automatic. It doesn’t.

Without a defined cadence, most people use the MCP connection reactively — they ask Claude something when they feel anxious about a goal, get an answer, and move on. This works for spot-checks but misses the pattern-recognition layer that makes AI planning help genuinely useful.

Patterns require consistent data. Consistent data requires a regular logging habit. A regular logging habit requires low friction at the point of data entry.

The MCP Planning Loop addresses all three by making the interaction so brief and structured that skipping it has a higher cognitive cost than doing it.

The Three Cadences

Cadence 1: The 2-Minute Daily Log

Frequency: Every workday, at the same time — typically end of day. MCP commands used: log_progress, occasionally complete_milestone. Purpose: Build a data record that makes weekly summaries meaningful.

The daily log is not a planning session. It’s a data entry habit.

You open Claude, tell it what happened on each active goal, and Claude logs it. Two minutes maximum. If it takes longer than two minutes, you’re reflecting when you should just be recording.

Template prompt:

End of day log. [Goal name 1]: [what happened, numeric if possible]. 
[Goal name 2]: [what happened]. [Goal name 3]: [what happened]. 
Log all of these in Beyond Time.

Example:

End of day log. Writing: 900 words written. Product launch: 
reviewed competitor pricing, no milestone complete yet. 
Fitness: skipped — evening call ran long. Log these in Beyond Time.

Claude will call log_progress for each goal mentioned and confirm what was recorded. The whole exchange takes under two minutes.

One important discipline: log on days when nothing happened. A zero entry is still data. A missing entry creates a gap in the record that the weekly summary can’t distinguish from a gap in your logging.

Cadence 2: The 20-Minute Weekly Review

Frequency: Once per week, consistent day and time — Friday afternoon or Sunday morning work well. MCP commands used: get_weekly_summary, list_milestones, add_milestone, sometimes create_goal or goal archival. Purpose: Identify patterns, surface stuck goals, and adjust plans.

The weekly review is where the MCP connection earns its setup cost. With five days of log entries, Claude can identify which goals have momentum, which are stalling, and what the data suggests about why.

We run the weekly review in four parts.

Part 1: Pull the summary.

Pull my weekly summary from Beyond Time. Show me which goals 
received entries and which didn't.

Part 2: Interpret the pattern.

Based on the summary, which goal concerns you most? What does 
the data suggest about the obstacle?

This is where Claude’s reasoning layer is most useful. It can notice that your health goal always gets skipped on Tuesdays and Thursdays (when you have back-to-back meetings) and flag the scheduling conflict explicitly.

Part 3: Milestone review.

For [goal nearest its target date], what milestones are still 
open? Are any overdue?

Part 4: Adjust or archive.

Is there any goal in my list that I should consider archiving 
based on this week's data? What's your reasoning?

This last prompt is deliberately challenging. Claude will sometimes push back and suggest a goal just needs a strategy change, not archival. That tension is productive — it forces you to make an explicit decision rather than let underperforming goals drag indefinitely.

Cadence 3: The 30-Minute Monthly Adjustment

Frequency: Once per month — first Sunday of the month, or last working day. MCP commands used: Full read across all goals, create_goal, archival. Purpose: Revise goal definitions, retire completed or abandoned goals, and set intentions for the next 30 days.

The monthly adjustment is a different kind of conversation than the weekly review. The weekly review asks “how did this week go?” The monthly adjustment asks “are we working on the right things?”

The prompt that anchors the session:

Pull all my active goals from Beyond Time. I want to do a 
monthly adjustment. For each goal, tell me: (1) the progress 
rate over the past 30 days, (2) whether that rate puts me on 
track for the target date, and (3) whether the goal definition 
still seems right given how the work has actually gone.

Claude will call list_goals and synthesize across the log data, returning a structured assessment of each goal. This gives you a decision-ready view of your portfolio rather than a vague sense of how things are going.

After reviewing, you make three types of decisions:

Continue as-is. Goals on track with clear milestones.

Revise. Goals where the target date or scope needs adjustment. The definition was right; the plan needs work.

Archive. Goals that haven’t received a log entry in 30+ days, or goals you’ve tacitly abandoned. Archiving isn’t failure — it’s maintaining signal clarity in your system.

The monthly adjustment typically reveals at least one goal that should be archived. Letting that goal sit in your active list creates background cognitive noise every time Claude summarizes your status.

The Data Quality Problem (and How to Avoid It)

The MCP Planning Loop produces better results when the underlying data is honest. Two patterns degrade data quality.

Optimistic logging. Logging 2 hours of “focused work” when the actual deep work time was 40 minutes surrounded by interruptions. Claude will treat the log entry at face value. If you log optimistically consistently, the weekly summary will look better than your results warrant.

The fix: log what actually happened, including context. “2 hours at desk, ~45 minutes actual focus” is more useful data than “2 hours working.”

Vanity milestones. Setting milestones that are easy to complete, then completing them to generate positive summary data. The weekly review looks great; the goal doesn’t move.

The fix: milestones should represent genuine progress gates. If completing a milestone doesn’t materially advance the goal, it probably shouldn’t be a milestone.

Neither of these requires perfect discipline. But being honest in your logs — even briefly noting when a session went poorly — is what separates a useful data record from a feel-good log.

How to Handle Goal Overlap

Most people have goals that compete for the same hours. A writing goal and a product launch goal both want morning deep work time. The MCP doesn’t resolve that conflict automatically — but it can surface it.

During the weekly review, add this prompt:

Looking at my goals, do you see any that are likely competing 
for the same type of time or energy? If so, which one should 
get priority this week and why?

Claude will reason over your goal categories, recent log patterns, and upcoming target dates to suggest a priority. This isn’t a decision you have to take, but it gives you an outside perspective before you set your week.

What Happens When the Framework Breaks Down?

The MCP Planning Loop is designed to be recoverable. Missing one daily log isn’t meaningful. Missing a weekly review isn’t a crisis. The system degrades gracefully.

The signal that something needs attention is three consecutive weeks with no log entries across any goal. At that point, the monthly adjustment question changes:

I haven't logged any goal progress in three weeks. What 
should we do first — review whether my goals are still 
relevant, or just restart the daily logging habit and see 
what sticks?

Sometimes the answer is to restart logging. Sometimes it’s to archive everything and start fresh with a smaller, more honest goal list. Claude can help you distinguish between a logging habit problem and a goal relevance problem based on what you tell it about your actual work.

Starting the Framework Today

You don’t need to implement all three cadences simultaneously. Start with just Cadence 1.

Do the 2-minute end-of-day log every workday for two weeks. By the end of the second week, you’ll have enough data to run a meaningful weekly review — and the friction of the review will be lower because the data is already there.

After the first proper weekly review, add Cadence 2. The monthly adjustment will follow naturally from having four weekly reviews to reflect on.

Your starting action: Tonight, open Claude Desktop and log today’s progress against your active Beyond Time goals. Keep it under two minutes. That’s the whole first step.

Related:

• The Complete Guide to Beyond Time MCP
• How to Set Up Beyond Time MCP
• 5 Prompts for Beyond Time via MCP
• Connecting AI Tools to Goals

Tags: beyond time MCP framework, AI planning loop, goal tracking cadence, claude planning system, MCP productivity