nvidia/optimization-from-data-orchestrator
cuopt_on_nemoclaw/openclaw-skills/optimization-from-data-orchestrator/SKILL.md
Use when a user uploads or provides data and asks a question that may be answered by optimization. This skill sequences optimization-intent-router, optimization-mode-router, tabular-optimization-ingestion, formulation skills, and cuOpt model-building skills.
$ install --global
skillsauthnpx skillsauth add nvidia/cuopt-examples optimization-from-data-orchestratorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 28, 2026, 6:19 AM184.8s1 file scanned
SKILL.md
- name:
- optimization-from-data-orchestrator
- summary:
- Coordinate the fast-path workflow for turning uploaded data and a natural-language question into the right optimization interpretation, clarification, cuOpt solve, and user-facing answer.
- description:
- Use when a user uploads or provides data and asks a question that may be answered by optimization. This skill sequences optimization-intent-router, optimization-mode-router, tabular-optimization-ingestion, formulation skills, and cuOpt model-building skills.
- origin:
- skill-evolution
Optimization From Data Orchestrator
Top-level coordinator for the fast path when a user provides data and asks a question that may be optimization. Sequences the supporting skills so the agent does not jump straight from uploaded data into a solver call.
When to use
All three must hold:
- the user has provided or is expected to provide data
- the question may be asking for the best / optimal / minimum / maximum decision under constraints
- the request is not yet so fully specified that you can call the solver directly
Skip this skill when the user is clearly asking for non-optimization analytics, the optimization problem is already fully specified mathematically, or the user has already chosen a dedicated replayable/auditable path.
Sequence
Run these in order, but skip any step already settled from context. Default to fast mode; surface replayable/auditable mode only on a real signal (reruns, audit, export, recurring planning).
optimization-intent-router— decide whether this is optimization at all and which family (LP / MILP / QP / routing). If non-optimization, stop the optimization flow.optimization-mode-router— only if there is a signal that replayability, audit, export, or recurring runs may matter. Otherwise stay in fast mode silently.tabular-optimization-ingestion— identify row grain and table roles, infer likely objective and constraint fields, refine the family classification if the data clearly supports a different one, and surface any blockers.cuopt-model-mapper— ask at most the final blocking clarification, then map directly into cuOpt and solve.
Family-specific handoffs after step 4:
- LP / MILP / QP →
numerical-optimization-formulationthencuopt-numerical-optimization-api-python(orcuopt-numerical-optimization-api-clifor MPS inputs) - Routing →
routing-formulationthencuopt-routing-api-python
Guardrails
- Do not skip intent classification and jump directly to cuOpt from raw data.
- Do not ask a long questionnaire before inspecting the uploaded data.
- Do not trigger replayable/auditable mode by default — only when the user signals reuse, audit, export, or recurring runs.
- Do not let ingestion become solver construction; the steps stay distinct.
- Do not use cuOpt for descriptive analytics tasks.
Related Skills
nvidia/tabular-optimization-ingestion
data-ai
VerifiedTrustedCommunity
Use when a user provides CSV, Excel, JSON-like tables, or similar structured data and asks a question that may become an LP, MILP, QP, or routing problem.
451SKILL.mdUpdated May 27, 2026
nvidia/optimization-mode-router
testing
VerifiedTrustedCommunity
Use when a user asks a question that may be answered by solving an optimization problem from uploaded or provided data, and you need to decide whether to proceed directly to cuOpt or preserve a structured reusable model artifact.
451SKILL.mdUpdated May 27, 2026
nvidia/optimization-intent-router
data-ai
VerifiedTrustedCommunity
Use when a user provides data and asks a natural-language business or planning question that may require optimization rather than simple analytics.
451SKILL.mdUpdated May 27, 2026
nvidia/cuopt-model-mapper
development
VerifiedTrustedCommunity
Use after optimization intent and basic data interpretation are established, when the goal is to solve quickly by mapping data directly into cuOpt rather than building a replayable intermediate artifact.
451SKILL.mdUpdated May 27, 2026