name	last-30-minutes
allowed-tools	mcp__memorylane__browse_timeline, mcp__memorylane__search_context, mcp__memorylane__get_activity_details
description	Summarize what you've been doing in the last 30 minutes.

Recent Activity

Summarize the user's recent screen activity.

Instructions

Step 1 — Fetch Recent Activity

Call browse_timeline(startTime="30 minutes ago", endTime="now", limit=50, sampling="recent_first").

Step 2 — Handle Empty Results

If nothing returned: Widen the window to browse_timeline(startTime="2 hours ago", endTime="now", limit=50, sampling="recent_first").
If the wider window also returns nothing: Report that MemoryLane may not be capturing (the app might not be running, or there has been no screen activity). Stop here.
If the wider window returns results but the 30-minute window didn't: Note the gap to the user (e.g., "No activity in the last 30 minutes, but here's what you were doing earlier").

Step 3 — Group and Summarize

From the returned entries:

Identify distinct apps from the activity summaries.
Cluster entries by task — use the summary text to group related activities together.
Estimate approximate time spent on each cluster from the timestamps of its entries.
Order groups by recency (most recent first).

Step 4 — Drill Into Details (only if needed)

Only call get_activity_details(ids) when a summary is ambiguous and the exact on-screen text would genuinely help clarify what the user was doing. Do not fetch OCR speculatively.

Step 5 — Present the Summary

Format as a brief narrative followed by bullet points:

**Last 30 minutes** (N activities recorded)

You were primarily working on [main task].

- **[App Name]** (~X min) — [what you were doing]
- **[App Name]** (~X min) — [what you were doing]

If the results came from the wider 2-hour window, adjust the heading accordingly.

Notes

Summaries are the primary source of truth. They are pre-generated from the captured activity and are sufficient for most reporting. Reserve get_activity_details for ambiguous cases.
Use search_context(query) if the user asks follow-up questions like "what was I doing in Chrome?" or "find that thing I was reading about X".
recent_first sampling is used instead of uniform because the user cares most about what just happened, and the window is short enough that uniform sampling would not add value.

Plus depuis ce dépôt

même dépôt

automation-instructions

deusXmachina-dev/memorylane

Write step-by-step automation instructions for a workflow, tailored to your tool (Claude, n8n or Zapier). Best run right after discover-patterns or process-analyst.

2026-06-04122

process-analyst-new

deusXmachina-dev/memorylane

(new) Find a person's repeated tasks from their screen activity, and what each one is worth automating, with the time and money saved. Outputs the numbers and data a report is built from, not the visual. Use to analyze processes, mine workflows, or see what is automatable and what it saves. Every figure is labelled and grounded, never made up.

2026-06-04122

report-new

deusXmachina-dev/memorylane

(new) Turn a process analysis into a polished, client-ready report, first an HTML deck you review, then a matching PDF. Use to package an analysis into an exec report or PDF. Shows the deck for approval before making the PDF, and never makes up numbers.

2026-06-04122

time-report

deusXmachina-dev/memorylane

Generate a time report grouped by client/project

2026-06-03122

discover-patterns

deusXmachina-dev/memorylane

Discover repeated workflow patterns from screen activity and suggest automations

2026-06-03122

release

deusXmachina-dev/memorylane

Prepare a MemoryLane release by updating the version and release notes, then creating and pushing the tagged release commit that triggers CI. Use when the user asks to release, ship, publish, bump version, or cut a stable or prerelease version.

2026-04-14122

Recent Activity

Summarize the user's recent screen activity.

Instructions

Step 1 — Fetch Recent Activity

Call browse_timeline(startTime="30 minutes ago", endTime="now", limit=50, sampling="recent_first").

Step 2 — Handle Empty Results

If nothing returned: Widen the window to browse_timeline(startTime="2 hours ago", endTime="now", limit=50, sampling="recent_first").

If the wider window also returns nothing: Report that MemoryLane may not be capturing (the app might not be running, or there has been no screen activity). Stop here.

If the wider window returns results but the 30-minute window didn't: Note the gap to the user (e.g., "No activity in the last 30 minutes, but here's what you were doing earlier").

Step 3 — Group and Summarize

From the returned entries:

Identify distinct apps from the activity summaries.

Cluster entries by task — use the summary text to group related activities together.

Estimate approximate time spent on each cluster from the timestamps of its entries.

Order groups by recency (most recent first).

Step 4 — Drill Into Details (only if needed)

Only call get_activity_details(ids) when a summary is ambiguous and the exact on-screen text would genuinely help clarify what the user was doing. Do not fetch OCR speculatively.

Step 5 — Present the Summary

Format as a brief narrative followed by bullet points:

**Last 30 minutes** (N activities recorded) You were primarily working on [main task]. - **[App Name]** (~X min) — [what you were doing] - **[App Name]** (~X min) — [what you were doing]

If the results came from the wider 2-hour window, adjust the heading accordingly.

Notes

Summaries are the primary source of truth. They are pre-generated from the captured activity and are sufficient for most reporting. Reserve get_activity_details for ambiguous cases.

Use search_context(query) if the user asks follow-up questions like "what was I doing in Chrome?" or "find that thing I was reading about X".

recent_first sampling is used instead of uniform because the user cares most about what just happened, and the window is short enough that uniform sampling would not add value.