Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Before setting up dependencies, writing data models, or configuring security
.claude/skills/firebase-firestore/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -23% | 0% |
Before setting up dependencies, writing data models, or configuring security rules, you MUST always identify the Firestore instance edition.
Run the following command to list current Firestore databases: bash npx -y firebase-tools@latest firestore:databases:list
bash npx -y firebase-tools@latest firestore:databases:get <database-id>
prefer to create a new instance.
edition is STANDARD, follow the guides underreferences/standard/.
edition is ENTERPRISE or native mode, follow the guidesunder references/enterprise/.
If no databases exist or the user requests a new one, default to provisioning an Enterprise edition database and ask the user what location to use. Run npx -y firebase-tools@latest firestore:locations to get the list of options. Suggest colocating with other resources if applicable.
Once the location is determined, create the database: bash npx -y firebase-tools@latest firestore:databases:create <database-id> --edition="enterprise" --location="<selected-location>"
Proceed with using the guides under references/enterprise/.
______________________________________________________________________
Based on the identified or created instance edition, open and read the corresponding reference guides:
references/standard/)firestore-rules-creation skillandroid_sdk_usage.md, ios_setup.md, or flutter_setup.md
references/enterprise/)provisioning.md
firestore-rules-creation skill> !CRITICAL] Mandatory Reference Reading Before writing or modifying any > application code for Firestore Enterprise Edition, you MUST read at > least one of the relevant reference documents below for the target > platform/language to understand specific architectural requirements and > pipeline initialization patterns.
Read web_sdk_usage.md, python_sdk_usage.md, android_sdk_usage.md, ios_setup.md, or flutter_setup.md
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | fail→pass | 18,117 | 7,185 | -60% | 1 | 1 | 0% | 1,718 | 1,048 | -39% | 0 | 0 | — |
case-01 | fail→fail | 12,719 | 8,548 | -33% | 1 | 1 | 0% | 558 | 1,007 | +80% | 0 | 0 | — |
case-02 | fail→fail | 20,543 | 13,569 | -34% | 1 | 1 | 0% | 2,599 | 1,090 | -58% | 0 | 0 | — |
case-03 | fail→pass | 20,219 | 8,923 | -56% | 1 | 1 | 0% | 2,350 | 1,307 | -44% | 0 | 0 | — |
case-04 | fail→pass | 12,036 | 11,961 | -1% | 1 | 1 | 0% | 1,607 | 1,698 | +6% | 0 | 0 | — |
case-05 | fail→pass | 32,547 | 8,451 | -74% | 1 | 1 | 0% | 2,830 | 1,348 | -52% | 0 | 0 | — |
case-06 | fail→pass | 16,992 | 3,678 | -78% | 1 | 1 | 0% | 1,786 | 1,383 | -23% | 0 | 0 | — |
case-07 | fail→fail | 18,344 | 8,034 | -56% | 1 | 1 | 0% | 1,912 | 1,156 | -40% | 0 | 0 | — |
case-08 | fail→pass | 17,506 | 8,714 | -50% | 1 | 1 | 0% | 2,523 | 1,178 | -53% | 0 | 0 | — |
case-09 | fail→fail | 13,848 | 8,020 | -42% | 1 | 1 | 0% | 2,251 | 1,155 | -49% | 0 | 0 | — |
case-10 | fail→fail | 19,923 | 11,266 | -43% | 1 | 1 | 0% | 2,452 | 2,149 | -12% | 0 | 0 | — |
case-11 | fail→pass | 9,803 | 11,024 | +12% | 1 | 1 | 0% | 1,445 | 1,708 | +18% | 0 | 0 | — |
case-12 | pass→pass | 19,210 | 20,956 | +9% | 1 | 1 | 0% | 2,579 | 3,044 | +18% | 0 | 0 | — |
case-13 | pass→pass | 16,076 | 17,566 | +9% | 1 | 1 | 0% | 1,758 | 2,827 | +61% | 0 | 0 | — |
case-14 | pass→pass | 24,185 | 24,138 | -0% | 1 | 1 | 0% | 2,816 | 3,449 | +22% | 0 | 0 | — |
case-15 | fail→fail | 14,763 | 21,300 | +44% | 1 | 1 | 0% | 1,647 | 2,455 | +49% | 0 | 0 | — |
case-16 | fail→pass | 14,148 | 9,603 | -32% | 1 | 1 | 0% | 1,252 | 1,530 | +22% | 0 | 0 | — |
case-17 | fail→pass | 16,886 | 7,552 | -55% | 1 | 1 | 0% | 1,691 | 1,153 | -32% | 0 | 0 | — |
case-18 | fail→fail | 19,434 | 7,783 | -60% | 1 | 1 | 0% | 1,744 | 1,114 | -36% | 0 | 0 | — |
case-20 | fail→fail | 8,887 | 7,337 | -17% | 1 | 1 | 0% | 1,289 | 1,052 | -18% | 0 | 0 | — |
case-21 | fail→pass | 12,804 | 15,539 | +21% | 1 | 1 | 0% | 1,838 | 1,136 | -38% | 0 | 0 | — |
case-22 | fail→pass | 21,728 | 9,916 | -54% | 1 | 1 | 0% | 1,574 | 1,505 | -4% | 0 | 0 | — |
case-23 | fail→pass | 15,147 | 17,006 | +12% | 1 | 1 | 0% | 1,460 | 1,665 | +14% | 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. 23 cases were attempted. The headline lift of +52 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/2/2026 | +68% |
| gemini-3.6-flash | verified | 7/23/2026 | +68% |
Other measured skills in the registry, with their headline benchmark lift.