Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Answer a question by querying the user's llmwiki. Use when the user asks about their own past work — "what did I decide about X", "what have I been working on", "how did I solve Y", "what's my preferred approach to Z", or any question that the wiki (built from their session history) might answer. Always read the wiki first before falling back to external search.
.claude/skills/pratiyush-llmwiki-query/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | -41% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-10 | ✓→✗ | ▼ Worse | -50% | 0% |
| case-12 | ✓→✗ | ▼ Worse | 173% | 0% |
Reads the user's local llmwiki (llmwiki/wiki/) and synthesises an answer to their question, with [[wikilink]] citations pointing to the source pages.
Invoke this skill when the user asks a question about their own work or history. Examples:
Do NOT invoke for questions that can be answered without the user's personal knowledge base (e.g., "how does TCP work?" → answer from general knowledge).
Follow the Query Workflow from the repo's CLAUDE.md:
llmwiki-sync skill for the fallback search order).wiki/index.md to identify the set of pages that might be relevant.wiki/overview.md to get the living synthesis context.[[wikilink]] citations. Quote directly from source pages when the wording matters.markdown Would you like me to save this as wiki/syntheses/<slug>.md for future reference? If yes, write it using the Synthesis Page Format from CLAUDE.md.
wiki/log.md: ## [YYYY-MM-DD] query | <short question>
If the wiki has no relevant pages:
/wiki-sync to pull in any recent sessions that might cover the topic.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,980 | 15,297 | +207% | 1 | 1 | 0% | 689 | 1,009 | +46% | 0 | 0 | — |
case-02 | fail→fail | 2,717 | 5,402 | +99% | 1 | 1 | 0% | 378 | 958 | +153% | 0 | 0 | — |
case-03 | fail→fail | 3,468 | 4,666 | +35% | 1 | 1 | 0% | 509 | 808 | +59% | 0 | 0 | — |
case-04 | fail→fail | 3,674 | 4,973 | +35% | 1 | 1 | 0% | 446 | 844 | +89% | 0 | 0 | — |
case-05 | fail→fail | 4,640 | 4,638 | -0% | 1 | 1 | 0% | 706 | 843 | +19% | 0 | 0 | — |
case-06 | fail→fail | 3,765 | 5,488 | +46% | 1 | 1 | 0% | 487 | 861 | +77% | 0 | 0 | — |
case-07 | fail→fail | 13,852 | 4,790 | -65% | 1 | 1 | 0% | 2,009 | 959 | -52% | 0 | 0 | — |
case-08 | fail→fail | 2,895 | 4,710 | +63% | 1 | 1 | 0% | 373 | 802 | +115% | 0 | 0 | — |
case-09 | fail→fail | 2,729 | 5,113 | +87% | 1 | 1 | 0% | 369 | 902 | +144% | 0 | 0 | — |
case-10 | pass→fail | 10,670 | 5,880 | -45% | 1 | 1 | 0% | 1,738 | 872 | -50% | 0 | 0 | — |
case-11 | fail→fail | 3,048 | 4,022 | +32% | 1 | 1 | 0% | 487 | 850 | +75% | 0 | 0 | — |
case-12 | pass→fail | 2,610 | 17,508 | +571% | 1 | 1 | 0% | 310 | 846 | +173% | 0 | 0 | — |
case-13 | pass→pass | 5,311 | 2,196 | -59% | 1 | 1 | 0% | 766 | 914 | +19% | 0 | 0 | — |
case-14 | fail→pass | 9,854 | 1,753 | -82% | 1 | 1 | 0% | 1,489 | 872 | -41% | 0 | 0 | — |
case-15 | fail→fail | 12,681 | 5,849 | -54% | 1 | 1 | 0% | 1,749 | 838 | -52% | 0 | 0 | — |
case-16 | fail→pass | 19,412 | 2,985 | -85% | 1 | 1 | 0% | 1,144 | 1,125 | -2% | 0 | 0 | — |
case-17 | pass→pass | 8,242 | 2,391 | -71% | 1 | 1 | 0% | 1,229 | 981 | -20% | 0 | 0 | — |
case-18 | pass→pass | 6,869 | 6,090 | -11% | 1 | 1 | 0% | 1,008 | 1,638 | +63% | 0 | 0 | — |
case-19 | fail→pass | 10,669 | 2,427 | -77% | 1 | 1 | 0% | 1,601 | 1,005 | -37% | 0 | 0 | — |
case-20 | pass→pass | 19,543 | 7,564 | -61% | 1 | 1 | 0% | 2,549 | 1,912 | -25% | 0 | 0 | — |
case-21 | pass→pass | 9,743 | 4,574 | -53% | 1 | 1 | 0% | 1,782 | 1,395 | -22% | 0 | 0 | — |
case-22 | pass→pass | 13,257 | 9,189 | -31% | 1 | 1 | 0% | 2,120 | 2,102 | -1% | 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, and 8 counted toward the lift figure. The other 14 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +5 percentage points is the difference between those two pass rates over the 8 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
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.