Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Takes raw input (text or file), classifies it along 5 Signal dimensions (Mode, Genre, Type, Format, Structure), routes it to the correct node(s), writes signal files, indexes in SQLite with FTS5, records in episodic memory, and feeds SICA learning patterns. Rejects signals with S/N < 0.3.
.claude/skills/miosa-osa-ingest/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 103% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -40% | 0% |
Ingest a signal into the OptimalOS knowledge base.
Auto-classifies S=(M,G,T,F,W), routes to node(s), writes signal file, indexes in SQLite FTS5, extracts entities, builds graph edges, records in episodic memory. Rejects signals with S/N < 0.3.
/ingest "Ed called about pricing, wants $2K per seat"
/ingest "Sales team closed the enterprise deal" --genre note
/ingest --file path/to/notes.md --genre transcript --title "Team Sync"| Flag | Description | Default | |------|-------------|---------| | --genre | Force genre (note, transcript, decision-log, etc.) | Auto-detect | | --title | Override auto-generated title | From content | | --file | Ingest from file instead of inline text | — | | --node | Force routing to specific node | Auto-route |
bashcd engine && mix optimal.ingest "content" --genre note cd engine && mix optimal.ingest --file path.md --genre transcript --title "Title"
The engine automatically handles indexing and cross-referencing. You still need to:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 4,612 | 4,962 | +8% | 1 | 1 | 0% | 651 | 1,047 | +61% | 0 | 0 | — |
case-02 | fail→fail | 25,148 | 8,905 | -65% | 1 | 1 | 0% | 4,355 | 1,012 | -77% | 0 | 0 | — |
case-03 | fail→pass | 5,499 | 6,461 | +17% | 1 | 1 | 0% | 714 | 1,306 | +83% | 0 | 0 | — |
case-04 | pass→fail | 8,143 | 6,384 | -22% | 1 | 1 | 0% | 1,305 | 1,282 | -2% | 0 | 0 | — |
case-05 | fail→fail | 6,133 | 12,297 | +101% | 1 | 1 | 0% | 843 | 1,735 | +106% | 0 | 0 | — |
case-06 | fail→fail | 2,686 | 5,564 | +107% | 1 | 1 | 0% | 401 | 659 | +64% | 0 | 0 | — |
case-07 | fail→pass | 7,779 | 4,259 | -45% | 1 | 1 | 0% | 555 | 1,129 | +103% | 0 | 0 | — |
case-08 | fail→pass | 10,307 | 3,204 | -69% | 1 | 1 | 0% | 1,430 | 857 | -40% | 0 | 0 | — |
case-09 | pass→pass | 11,222 | 2,353 | -79% | 1 | 1 | 0% | 1,770 | 716 | -60% | 0 | 0 | — |
case-10 | fail→pass | 7,260 | 2,433 | -66% | 1 | 1 | 0% | 1,175 | 708 | -40% | 0 | 0 | — |
case-11 | fail→pass | 5,179 | 2,926 | -44% | 1 | 1 | 0% | 786 | 866 | +10% | 0 | 0 | — |
case-12 | fail→pass | 9,118 | 7,091 | -22% | 1 | 1 | 0% | 1,627 | 1,658 | +2% | 0 | 0 | — |
case-13 | fail→pass | 6,401 | 5,557 | -13% | 1 | 1 | 0% | 976 | 814 | -17% | 0 | 0 | — |
case-14 | fail→pass | 1,801 | 4,688 | +160% | 1 | 1 | 0% | 244 | 1,198 | +391% | 0 | 0 | — |
case-15 | fail→pass | 16,751 | 3,536 | -79% | 1 | 1 | 0% | 2,756 | 650 | -76% | 0 | 0 | — |
case-16 | pass→fail | 11,538 | 6,944 | -40% | 1 | 1 | 0% | 1,910 | 821 | -57% | 0 | 0 | — |
case-17 | pass→pass | 9,058 | 5,989 | -34% | 1 | 1 | 0% | 1,558 | 1,399 | -10% | 0 | 0 | — |
case-18 | fail→pass | 12,190 | 4,320 | -65% | 1 | 1 | 0% | 2,001 | 1,121 | -44% | 0 | 0 | — |
case-19 | fail→pass | 5,089 | 3,241 | -36% | 1 | 1 | 0% | 849 | 920 | +8% | 0 | 0 | — |
case-20 | fail→pass | 3,845 | 9,322 | +142% | 1 | 1 | 0% | 645 | 1,828 | +183% | 0 | 0 | — |
case-21 | fail→pass | 13,548 | 5,425 | -60% | 1 | 1 | 0% | 2,049 | 1,139 | -44% | 0 | 0 | — |
case-22 | fail→pass | 5,598 | 2,214 | -60% | 1 | 1 | 0% | 844 | 579 | -31% | 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 18 counted toward the lift figure. The other 4 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 +59 percentage points is the difference between those two pass rates over the 18 comparable cases. 3 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.