Palace walk

Skill frankxai/mind-palace-agent-skills/skills/palace-walk

Walk the palace from memory to self-test recall, score it, and repair weak loci. Use when the user says "test me on my palace", "let me recall the list", "walk my palace", "quiz me", or after a palace is built or due for review. Runs retrieval practice (the testing effect), marks which loci held and which failed, and updates the recall block of palace.json.From its SKILL.md

Install

npx -y skills add frankxai/mind-palace-agent-skills --skill palace-walk

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.4 KB, 523 tokens by cl100k_base, as published. Nobody here has run it

palace-walk

The retrieval step. Recalling is what makes memory durable — re-reading is not. Make the user walk the route and produce each item before showing them anything.

When to use this

A palace exists and the user wants to practise, self-test, or do a scheduled review. This is where learning actually consolidates.

Instructions

Walk in route order. Take rooms and stations in their fixed order from palace.json. At each station, prompt for recall before revealing anything: "Station: kitchen table — what's here?"
Test, don't show. Let the user attempt retrieval cold. Reveal the stored image.text and target only after their attempt. The effortful attempt is the active ingredient, even when it fails (productive failure).
Score each locus honestly: got (clean), partial (image but not target, or vice versa), missed. Record per-locus, not just a total.
Repair on the spot. For partial/missed, diagnose: weak image (→ re-encode via loci-encoder with stronger levers), crowded station (→ ask memory-palace-architect to add spacing), or just under-rehearsed (→ schedule sooner). Make the fix, then re-walk that locus.
Update recall state. Write each locus's recall block (or create it on first review): reps, lapses, lastReviewed, and hand the grades to spaced-recall to compute the next due date and stability/difficulty.
Report the walk. End with a terse line: items tested, got/partial/missed counts, and the loci queued for repair. No celebration — just the state.

Refusals

Never reveal the answer before the user attempts recall — that turns a test back into re-reading and kills the testing effect.
Don't mark a locus got on a partial; honest scoring is what makes the schedule work.

References

spec/MEMORY-PALACE-METHOD.md §4 (testing effect, durability).

Built on SIP · mind-palace-agent-skills · Memory Palace Method v0.1

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.