Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create agent rules.
.claude/skills/griddynamics-init-workspace-rules/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 642% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -21% | 0% |
<init_workspace_rules>
<role> Senior agent configuration specialist — Rosetta-to-local full-copy adaptation expert. </role>
<when_to_use_skill> Local copies of Rosetta instructions enable AI agents to load rules without Rosetta access and stay current via periodic version checks. Creates full local files for all Rosetta content adapted to detected IDE/CodingAgent format. Validation: all Rosetta content exists as local files, root entry point triggers full prep chain. </when_to_use_skill>
<core_concepts>
init-workspace-* skills/workflows, templates/shell-schemas/*, configure/*, rules/bootstrap.md MUST NOT BE copied<rosetta:file> sections, strip tags, merge into one file with one frontmatterinit creates all files; upgrade fills gaps only, never overwrites human-customized files</core_concepts>
<process>
Internal knowledge about IDE/agent configuration is obsolete — LIST and ACQUIRE from KB.
Step 1: Identify Environment
configure IN KB with XML format (to understand supported IDE/CodingAgents)Step 2: Read Workspace Context
Step 3: Discover Full Rosetta Content (subagent)
all FROM KB with format=flat, save to FEATURE TEMP folder as list-all-output.mdStep 4: MUST Install Root Entry Point and Bootstrap Rules
rules/local-files-mode.md FROM KB — install as root entry point per IDE configure specrules/bootstrap-*.md FROM KB — install as individual rule files per IDE configure specStep 5: MUST Generate All Content Files
For each content type from filtered list (non-bootstrap rules, skills, agents, workflows, commands):
Step 6: Verify and Report (HITL)
<rosetta:file> tags, single frontmatter per file)</process>
<validation_checklist>
</validation_checklist>
<pitfalls>
</pitfalls>
</init_workspace_rules>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 16,809 | 12,232 | -27% | 1 | 1 | 0% | 2,831 | 2,942 | +4% | 0 | 0 | — |
case-02 | fail→fail | 3,028 | 7,701 | +154% | 1 | 1 | 0% | 419 | 1,391 | +232% | 0 | 0 | — |
case-03 | fail→pass | 2,230 | 8,952 | +301% | 1 | 1 | 0% | 328 | 2,434 | +642% | 0 | 0 | — |
case-04 | pass→pass | 15,132 | 16,572 | +10% | 1 | 1 | 0% | 2,409 | 3,459 | +44% | 0 | 0 | — |
case-05 | fail→fail | 5,558 | 8,226 | +48% | 1 | 1 | 0% | 798 | 2,334 | +192% | 0 | 0 | — |
case-06 | pass→pass | 12,896 | 13,713 | +6% | 1 | 1 | 0% | 2,334 | 3,463 | +48% | 0 | 0 | — |
case-07 | fail→pass | 10,310 | 4,686 | -55% | 1 | 1 | 0% | 1,729 | 1,741 | +1% | 0 | 0 | — |
case-08 | fail→pass | 8,014 | 2,603 | -68% | 1 | 1 | 0% | 1,343 | 1,424 | +6% | 0 | 0 | — |
case-09 | fail→pass | 11,372 | 3,547 | -69% | 1 | 1 | 0% | 1,659 | 1,456 | -12% | 0 | 0 | — |
case-10 | fail→pass | 14,835 | 5,372 | -64% | 1 | 1 | 0% | 2,403 | 1,893 | -21% | 0 | 0 | — |
case-11 | pass→pass | 12,021 | 4,207 | -65% | 1 | 1 | 0% | 1,953 | 1,660 | -15% | 0 | 0 | — |
case-12 | fail→pass | 10,744 | 3,057 | -72% | 1 | 1 | 0% | 1,600 | 1,363 | -15% | 0 | 0 | — |
case-13 | pass→pass | 12,385 | 4,104 | -67% | 1 | 1 | 0% | 1,915 | 1,676 | -12% | 0 | 0 | — |
case-14 | fail→pass | 10,459 | 1,805 | -83% | 1 | 1 | 0% | 1,487 | 1,212 | -18% | 0 | 0 | — |
case-15 | pass→pass | 11,940 | 5,299 | -56% | 1 | 1 | 0% | 1,936 | 1,766 | -9% | 0 | 0 | — |
case-16 | pass→pass | 8,066 | 6,056 | -25% | 1 | 1 | 0% | 1,123 | 1,964 | +75% | 0 | 0 | — |
case-17 | pass→pass | 13,656 | 4,068 | -70% | 1 | 1 | 0% | 2,212 | 1,584 | -28% | 0 | 0 | — |
case-18 | pass→pass | 25,699 | 3,335 | -87% | 1 | 1 | 0% | 2,338 | 1,452 | -38% | 0 | 0 | — |
case-19 | pass→fail | 6,867 | 2,747 | -60% | 1 | 1 | 0% | 977 | 1,335 | +37% | 0 | 0 | — |
case-20 | fail→pass | 8,508 | 4,351 | -49% | 1 | 1 | 0% | 1,318 | 1,589 | +21% | 0 | 0 | — |
case-21 | fail→fail | 6,929 | 2,194 | -68% | 1 | 1 | 0% | 1,142 | 1,232 | +8% | 0 | 0 | — |
case-22 | pass→pass | 24,785 | 6,556 | -74% | 1 | 1 | 0% | 2,226 | 1,843 | -17% | 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. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 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.
Other measured skills in the registry, with their headline benchmark lift.