Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Designs memory palace structures with spatial layouts and domain organization. Use when creating a new palace or planning knowledge architecture by hand.
.claude/skills/athola-memory-palace-architect/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 123% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 66% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -48% | 0% |
> Status: unwired. As of v1.9.4, no command or agent > invokes Skill(memory-palace:memory-palace-architect). > The /palace create command does palace creation directly > without routing through this skill. The architectural > guidance below is reference material; consult it when > designing a new palace by hand. Tracked for follow-up in > the April 2026 skill audit.
Design and construct virtual memory palaces for spatial knowledge organization. This skill guides you through creating memorable spatial structures that enhance recall and organize complex information.
A memory palace is a mnemonic technique that uses spatial visualization to organize and recall information. This skill provides a systematic approach for:
bashpython scripts/palace_manager.py create "My Palace" "programming" --metaphor workshop
Verification: Run python --version to verify Python environment.
bashpython scripts/palace_manager.py list
Verification: Run python --version to verify Python environment.
bashpython scripts/palace_manager.py status
Verification: Run python --version to verify Python environment.
session-palace-builder
session-palace-builder
| Template | Best For | Key Features | |----------|----------|--------------| | Fortress | Security, defense, production-grade systems | Strong boundaries, layered access | | Library | Knowledge, research, documentation | Organized shelves, categorized sections | | Workshop | Practical skills, tools, techniques | Workbenches, tool areas, project spaces | | Garden | Organic growth, evolving knowledge | Plots, seasons, interconnected paths | | Observatory | Exploration, discovery, patterns | Viewing platforms, star maps, instruments |
modules/domain-analysis.mdmodules/layout-patterns.mdmodules/sensory-encoding.mdmodules/validation.mdmodules/franklin-protocol.md - Apply the original learning algorithm to palace designWorks with:
knowledge-locator - For searching across palacessession-palace-builder - For temporary session palacesdigital-garden-cultivator - For evolving knowledge basesIf palace creation fails, check that the metaphor argument matches one of the supported templates (Fortress, Library, Workshop, Garden, Observatory). For script errors, ensure the palace_manager.py script has executable permissions and that your Python environment meets the requirements listed in pyproject.toml.
<template> completes without error and adds the palace to the index
scripts/palace_manager.py list shows the newly created palaceby the name supplied during creation
encoding profile, and a navigation guide
metaphor argument resolves to one of the five supportedtemplates (Fortress, Library, Workshop, Garden, Observatory); if not, error is reported with the valid list
produced as part of the Expected Outputs section
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→fail | 6,667 | 5,762 | -14% | 1 | 1 | 0% | 999 | 2,014 | +102% | 0 | 0 | — |
case-01 | fail→fail | 19,645 | 13,108 | -33% | 1 | 1 | 0% | 3,181 | 3,377 | +6% | 0 | 0 | — |
case-02 | fail→pass | 5,918 | 6,599 | +12% | 1 | 1 | 0% | 1,038 | 2,316 | +123% | 0 | 0 | — |
case-03 | fail→fail | 35,232 | 33,222 | -6% | 1 | 1 | 0% | 6,202 | 6,483 | +5% | 0 | 0 | — |
case-04 | fail→pass | 7,176 | 5,071 | -29% | 1 | 1 | 0% | 1,069 | 1,925 | +80% | 0 | 0 | — |
case-06 | fail→pass | 30,466 | 12,174 | -60% | 1 | 1 | 0% | 3,011 | 3,283 | +9% | 0 | 0 | — |
case-07 | fail→fail | 10,328 | 4,182 | -60% | 1 | 1 | 0% | 1,609 | 1,851 | +15% | 0 | 0 | — |
case-08 | fail→fail | 9,994 | 3,116 | -69% | 1 | 1 | 0% | 1,597 | 1,772 | +11% | 0 | 0 | — |
case-09 | fail→fail | 12,087 | 3,601 | -70% | 1 | 1 | 0% | 1,797 | 1,691 | -6% | 0 | 0 | — |
case-10 | fail→pass | 6,400 | 4,015 | -37% | 1 | 1 | 0% | 1,062 | 1,766 | +66% | 0 | 0 | — |
case-11 | fail→pass | 19,655 | 2,117 | -89% | 1 | 1 | 0% | 2,836 | 1,486 | -48% | 0 | 0 | — |
case-12 | fail→pass | 15,595 | 2,129 | -86% | 1 | 1 | 0% | 2,280 | 1,512 | -34% | 0 | 0 | — |
case-13 | fail→pass | 8,923 | 2,056 | -77% | 1 | 1 | 0% | 1,211 | 1,477 | +22% | 0 | 0 | — |
case-14 | fail→pass | 14,073 | 2,299 | -84% | 1 | 1 | 0% | 1,935 | 1,388 | -28% | 0 | 0 | — |
case-15 | pass→pass | 8,153 | 2,785 | -66% | 1 | 1 | 0% | 1,195 | 1,531 | +28% | 0 | 0 | — |
case-16 | fail→pass | 8,924 | 2,626 | -71% | 1 | 1 | 0% | 1,434 | 1,472 | +3% | 0 | 0 | — |
case-17 | pass→pass | 13,452 | 5,127 | -62% | 1 | 1 | 0% | 2,142 | 1,897 | -11% | 0 | 0 | — |
case-18 | fail→pass | 15,543 | 3,004 | -81% | 1 | 1 | 0% | 2,253 | 1,609 | -29% | 0 | 0 | — |
case-19 | fail→pass | 10,077 | 1,496 | -85% | 1 | 1 | 0% | 1,424 | 1,346 | -5% | 0 | 0 | — |
case-20 | fail→pass | 7,203 | 3,717 | -48% | 1 | 1 | 0% | 1,121 | 1,417 | +26% | 0 | 0 | — |
case-21 | pass→pass | 10,748 | 2,133 | -80% | 1 | 1 | 0% | 1,819 | 1,367 | -25% | 0 | 0 | — |
case-22 | fail→pass | 12,202 | 2,770 | -77% | 1 | 1 | 0% | 1,915 | 1,544 | -19% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +59 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.