Install any skill in seconds. Free to start, no credit card required.
Get Started Free →silverstein/minutes-ideas
Surface recent voice memos and ideas captured from any device.
Passed 4 of 10 test cases with the skill, and 2 of 10 without it.
and 2 more in the full report
4 fixed4 passed either way13 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
I have an audio recording file at '/tmp/voice_note.m4a' that has 10 seconds of silence at the start. Can you use ffmpeg to trim off the first 10 seconds and save it as '/tmp/trimmed.m4a'?
| Checked | Without skill | With skill |
|---|---|---|
| Executes ffmpeg or an equivalent audio editing command to trim the audio file | Missed | Passed |
| Saves the trimmed file to the requested target path | Missed | Passed |
The instructions your model reads when the skill loads.
Surface voice memos and ideas captured from any device in the last 14 days. This is the recall layer for the cross-device ghost context pipeline.
minutes CLI:minutes list --content-type memo --limit 20
Require exit status 0 and use the JSON written to stdout. If the CLI is unavailable or fails, report the memo source unavailable; never scan the memo directory directly.
Present the memos as a clean list:
If the user picks one, run minutes get "<exact path>" --json, require exit
status 0, and present only the returned transcript/summary. Never reopen the
list path through the host filesystem.
These memos were captured on the user's phone (or Mac) and automatically transcribed by the Minutes watcher. They may contain ideas, thoughts, observations, or reminders that the user recorded while away from their desk.
When the user asks "what was that idea I had while walking?" — search these memos first, then broaden to full meeting search if needed.
test cases passed, vs 2 / 10 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.