Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Facilitates conversational discovery to create Architectural Decision Records (ADRs) for functional requirements covering CLI, REST/HTTP APIs, or both. Use when the user wants to document command-line or HTTP service architecture, capture functional requirements, create ADRs for CLI or API projects, or design interfaces with documented decisions. This should trigger for requests such as Create ADR for functional requirements; Document functional requirements; Capture functional requirements; Gen
.claude/skills/jabrena-031-architecture-adr-functional-requirements/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 6% | 0% |
Guide stakeholders through a structured conversation to uncover and document technical decisions and functional requirements for command-line tools, REST/HTTP APIs, or combined surfaces. This is an interactive SKILL. The ADR is the documentation of that conversation, not the conversation itself. Infer CLI vs API from the current workspace when possible; ask a short clarifying question when unclear. Use only the current conversation and repository files explicitly available in the current session.
What is covered in this Skill?
Use conversational discovery in small batches, build on answers, validate before proceeding. Only create ADR after thorough conversation and user confirmation.
date command before starting to get accurate timestamps for the ADRUse the local shell date command before discovery and use it for ADR timestamps.
Load references/031-architecture-adr-functional-requirements.md from this skill, infer CLI/API scope from the current workspace, and ask a short clarifying question if unclear.
Guide discovery in small batches to elicit functional requirements and technical decisions for CLI, REST API, or both.
Step constraints:
Only after user confirms proceed, generate the ADR document and provide concise next steps.
For detailed guidance, examples, and constraints, see references/031-architecture-adr-functional-requirements.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | fail→fail | 6,330 | 6,027 | -5% | 1 | 1 | 0% | 875 | 843 | -4% | 0 | 0 | — |
case-21 | fail→fail | 10,260 | 6,897 | -33% | 1 | 1 | 0% | 1,382 | 1,017 | -26% | 0 | 0 | — |
case-01 | fail→fail | 8,180 | 5,066 | -38% | 1 | 1 | 0% | 1,175 | 865 | -26% | 0 | 0 | — |
case-02 | fail→pass | 10,560 | 7,176 | -32% | 1 | 1 | 0% | 1,572 | 1,613 | +3% | 0 | 0 | — |
case-03 | fail→fail | 6,948 | 5,364 | -23% | 1 | 1 | 0% | 1,084 | 885 | -18% | 0 | 0 | — |
case-04 | pass→fail | 21,082 | 7,290 | -65% | 1 | 1 | 0% | 2,970 | 1,017 | -66% | 0 | 0 | — |
case-05 | pass→fail | 18,340 | 14,171 | -23% | 1 | 1 | 0% | 2,831 | 2,756 | -3% | 0 | 0 | — |
case-06 | pass→fail | 19,826 | 10,337 | -48% | 1 | 1 | 0% | 3,071 | 2,314 | -25% | 0 | 0 | — |
case-07 | fail→pass | 13,169 | 10,251 | -22% | 1 | 1 | 0% | 2,032 | 2,242 | +10% | 0 | 0 | — |
case-08 | fail→fail | 15,297 | 16,253 | +6% | 1 | 1 | 0% | 2,177 | 3,057 | +40% | 0 | 0 | — |
case-09 | fail→fail | 8,004 | 9,429 | +18% | 1 | 1 | 0% | 1,247 | 1,362 | +9% | 0 | 0 | — |
case-10 | fail→fail | 11,355 | 5,789 | -49% | 1 | 1 | 0% | 1,680 | 892 | -47% | 0 | 0 | — |
case-11 | fail→pass | 9,855 | 8,180 | -17% | 1 | 1 | 0% | 1,473 | 1,384 | -6% | 0 | 0 | — |
case-12 | fail→fail | 8,608 | 8,642 | +0% | 1 | 1 | 0% | 1,215 | 1,538 | +27% | 0 | 0 | — |
case-13 | fail→pass | 9,883 | 6,007 | -39% | 1 | 1 | 0% | 1,493 | 1,441 | -3% | 0 | 0 | — |
case-14 | pass→pass | 7,838 | 5,740 | -27% | 1 | 1 | 0% | 1,073 | 1,409 | +31% | 0 | 0 | — |
case-15 | fail→pass | 8,455 | 5,126 | -39% | 1 | 1 | 0% | 1,137 | 1,206 | +6% | 0 | 0 | — |
case-16 | fail→fail | 8,533 | 5,503 | -36% | 1 | 1 | 0% | 1,209 | 793 | -34% | 0 | 0 | — |
case-17 | fail→pass | 9,118 | 9,923 | +9% | 1 | 1 | 0% | 1,344 | 1,527 | +14% | 0 | 0 | — |
case-18 | fail→pass | 7,452 | 6,162 | -17% | 1 | 1 | 0% | 1,113 | 1,462 | +31% | 0 | 0 | — |
case-19 | fail→pass | 8,293 | 5,820 | -30% | 1 | 1 | 0% | 1,193 | 1,368 | +15% | 0 | 0 | — |
case-20 | fail→fail | 8,114 | 5,388 | -34% | 1 | 1 | 0% | 1,213 | 814 | -33% | 0 | 0 | — |
case-23 | fail→pass | 11,503 | 6,343 | -45% | 1 | 1 | 0% | 1,778 | 1,423 | -20% | 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, and 14 counted toward the lift figure. The other 9 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 +26 percentage points is the difference between those two pass rates over the 14 comparable cases. 5 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.