Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the agent needs access to information beyond its training data — knowledge sources, RAG pipelines, or grounding data.
.claude/skills/sharpdeveye-enrich/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-10 | ✓→✗ | ▼ Worse | -10% | 0% |
| case-16 | ✓→✗ | ▼ Worse | -55% | 0% |
Invoke /agent-workflow — it contains workflow principles, anti-patterns, and the Context Gathering Protocol. Follow the protocol before proceeding — if no workflow context exists yet, you MUST run /teach-maestro first. Consult the knowledge-systems reference in the agent-workflow skill for RAG architecture, chunking strategies, and retrieval patterns.
Add knowledge sources to ground the workflow in facts. Without grounding, agents hallucinate. With grounding, they cite sources.
Identify what knowledge the workflow needs:
| Knowledge Type | Source | Update Frequency | Access Pattern | |---------------|--------|-----------------|----------------| | Domain docs | Internal docs, specs | Monthly | Semantic search | | Code context | Codebase | Real-time | Code search | | User data | Database, CRM | Real-time | Structured query | | External data | APIs, web | Real-time | API call | | Historical | Logs, past interactions | Daily | Time-range query |
For document-based knowledge (consult the knowledge-systems reference in the agent-workflow skill):
For database-backed knowledge:
For live information:
After enrichment, run /evaluate to test retrieval quality, or /iterate to set up continuous monitoring of knowledge freshness.
NEVER:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 25,153 | 25,597 | +2% | 1 | 1 | 0% | 4,131 | 4,747 | +15% | 0 | 0 | — |
case-02 | pass→pass | 28,669 | 28,083 | -2% | 1 | 1 | 0% | 4,875 | 5,683 | +17% | 0 | 0 | — |
case-03 | pass→pass | 23,789 | 21,148 | -11% | 1 | 1 | 0% | 3,718 | 4,234 | +14% | 0 | 0 | — |
case-04 | pass→pass | 4,946 | 6,153 | +24% | 1 | 1 | 0% | 758 | 1,800 | +137% | 0 | 0 | — |
case-13 | pass→pass | 13,319 | 13,892 | +4% | 1 | 1 | 0% | 2,123 | 2,865 | +35% | 0 | 0 | — |
case-05 | pass→pass | 4,510 | 5,924 | +31% | 1 | 1 | 0% | 826 | 1,619 | +96% | 0 | 0 | — |
case-06 | pass→pass | 3,058 | 4,419 | +45% | 1 | 1 | 0% | 559 | 1,420 | +154% | 0 | 0 | — |
case-07 | fail→pass | 17,820 | 15,073 | -15% | 1 | 1 | 0% | 2,866 | 3,351 | +17% | 0 | 0 | — |
case-08 | pass→pass | 13,238 | 18,617 | +41% | 1 | 1 | 0% | 2,095 | 3,612 | +72% | 0 | 0 | — |
case-09 | pass→pass | 15,695 | 8,209 | -48% | 1 | 1 | 0% | 1,705 | 1,822 | +7% | 0 | 0 | — |
case-10 | pass→fail | 10,595 | 4,261 | -60% | 1 | 1 | 0% | 1,575 | 1,413 | -10% | 0 | 0 | — |
case-11 | pass→pass | 14,517 | 8,714 | -40% | 1 | 1 | 0% | 2,145 | 1,932 | -10% | 0 | 0 | — |
case-12 | pass→pass | 15,422 | 14,539 | -6% | 1 | 1 | 0% | 2,451 | 2,790 | +14% | 0 | 0 | — |
case-14 | pass→pass | 13,411 | 11,400 | -15% | 1 | 1 | 0% | 2,168 | 2,412 | +11% | 0 | 0 | — |
case-15 | pass→pass | 8,853 | 4,983 | -44% | 1 | 1 | 0% | 1,266 | 1,385 | +9% | 0 | 0 | — |
case-16 | pass→fail | 17,454 | 6,980 | -60% | 1 | 1 | 0% | 2,547 | 1,143 | -55% | 0 | 0 | — |
case-17 | pass→pass | 16,774 | 12,109 | -28% | 1 | 1 | 0% | 2,146 | 2,613 | +22% | 0 | 0 | — |
case-18 | fail→pass | 11,375 | 10,371 | -9% | 1 | 1 | 0% | 1,768 | 2,161 | +22% | 0 | 0 | — |
case-19 | pass→pass | 11,696 | 11,339 | -3% | 1 | 1 | 0% | 1,867 | 2,430 | +30% | 0 | 0 | — |
case-20 | pass→pass | 13,016 | 11,033 | -15% | 1 | 1 | 0% | 1,991 | 2,354 | +18% | 0 | 0 | — |
case-21 | pass→pass | 14,930 | 12,997 | -13% | 1 | 1 | 0% | 2,333 | 2,858 | +23% | 0 | 0 | — |
case-22 | pass→pass | 16,881 | 11,345 | -33% | 1 | 1 | 0% | 1,636 | 2,356 | +44% | 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 21 counted toward the lift figure. The other 1 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 21 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.