Merge branch 'main' into fix/issue-1488-cpufj-incumbent

Author: aliceb-nv

skills/cuopt-numerical-optimization-api-cli/BENCHMARK.md

@@ -7,11 +7,11 @@ This benchmark summarizes 3-Tier Evaluation from NVSkills-Eval results for the s
 ## Evaluation Summary
 
 - Skill: `cuopt-numerical-optimization-api-cli`
-- Evaluation date: 2026-05-29
+- Evaluation date: 2026-07-01
 - NVSkills-Eval profile: `external`
-- Environment: `local`
+- Environment: `astra-sandbox`
 - Dataset: 1 evaluation tasks
-- Attempts per task: 2
+- Attempts per task: 1
 - Pass threshold: 50%
 - Overall verdict: PASS
 
@@ -54,34 +54,26 @@ Task composition is derived from the evaluation dataset when possible. Entries w
 
 | Dimension | Num | `claude-code` | `codex` |
 |---|---:|---:|---:|
-| Security | 2 | 100% (+0%) | 100% (+0%) |
-| Correctness | 2 | 100% (+0%) | 97% (+5%) |
-| Discoverability | 2 | 100% (+0%) | 84% (+5%) |
-| Effectiveness | 2 | 78% (+2%) | 76% (+4%) |
-| Efficiency | 2 | 93% (-0%) | 78% (-0%) |
+| Security | 1 | 100% (+0%) | 100% (+0%) |
+| Correctness | 1 | 30% (+0%) | 77% (+47%) |
+| Discoverability | 1 | 0% (+0%) | 84% (+84%) |
+| Effectiveness | 1 | 39% (-4%) | 41% (+4%) |
+| Efficiency | 1 | 27% (-0%) | 77% (+49%) |
 
 Score values show skill-assisted performance. Values in parentheses show uplift versus the no-skill baseline when baseline data is available.
 
 ## Tier 1: Static Validation Summary
 
-Tier 1 validation passed with observations. NVSkills-Eval ran 9 checks and found 8 total findings.
+Tier 1 validation passed with observations. NVSkills-Eval ran 1 checks and found 2 total findings.
 
 Top findings:
 
 - MEDIUM SCHEMA/body_recommended_section: Missing recommended section: '## Instructions' (`skills/cuopt-numerical-optimization-api-cli/SKILL.md`)
-- LOW QUALITY/quality_discoverability: Broad description without negative triggers may cause over-triggering (`skills/cuopt-numerical-optimization-api-cli/SKILL.md`)
-- LOW QUALITY/quality_discoverability: No '## Purpose' section (`skills/cuopt-numerical-optimization-api-cli/SKILL.md`)
-- LOW QUALITY/quality_reliability: No prerequisites/requirements documented (`skills/cuopt-numerical-optimization-api-cli/SKILL.md`)
-- LOW QUALITY/quality_reliability: No limitations documented (`skills/cuopt-numerical-optimization-api-cli/SKILL.md`)
+- LOW SCHEMA/author_format: Author must be of the form 'Name <email@host>' (`skills/cuopt-numerical-optimization-api-cli/SKILL.md`)
 
 ## Tier 2: Deduplication Summary
 
-Tier 2 validation passed. NVSkills-Eval ran 2 checks and found 0 total findings.
-
-Notable observations:
-
-- Context Deduplication: Collected 5 file(s)
-- Inter-Skill Deduplication: Parsed skill 'cuopt-numerical-optimization-api-cli': 141 char description
+This tier was not run or did not produce findings in this report.
 
 ## Publication Recommendation
 

skills/cuopt-numerical-optimization-api-cli/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: cuopt-numerical-optimization-api-cli
 version: "26.08.00"
-description: LP, MILP, and QP (beta) with cuOpt — CLI only (MPS files, cuopt_cli). Use when the user is solving LP, MILP, or QP from MPS via command line.
+description: LP, MILP, and QP (beta) with cuOpt — CLI only (MPS/LP/QPS files, cuopt_cli). Use when the user is solving LP, MILP, or QP from an MPS or LP file via command line.
 license: Apache-2.0
 metadata:
   author: NVIDIA cuOpt Team
@@ -17,63 +17,53 @@ metadata:
 
 # cuOpt Numerical Optimization — CLI
 
-Solve LP, MILP, and QP problems from MPS files via `cuopt_cli`. The same command, options, and MPS workflow apply across all three; QP uses the standard MPS quadratic-objective extension.
+Solve LP, MILP, and QP problems from MPS or LP files via `cuopt_cli`. The same command and options apply across all three; QP is supported in both MPS (QPS) and LP files.
 
 Confirm problem type and formulation (variables, objective, constraints, variable types) before coding.
 
-This skill is **CLI only** (MPS input).
+This skill is **CLI only** (MPS or LP file input).
 
 ## Basic usage
 
 ```bash
-# Solve LP or MILP from MPS file
-cuopt_cli problem.mps
-
-# With options
-cuopt_cli problem.mps --time-limit 120 --mip-relative-tolerance 0.01
+cuopt_cli <problem-file> [options]
 ```
 
-## Common options
+The first positional argument is the input file. The format is chosen automatically from the extension — MPS, QPS, and LP files are all accepted (including `.gz` / `.bz2` compressed variants); run `cuopt_cli --help` for the exact list of supported extensions.
 
-```bash
-cuopt_cli --help
+## Options
 
-# Time limit (seconds)
-cuopt_cli problem.mps --time-limit 120
+**`cuopt_cli --help` is the authoritative list — don't work from a hard-coded subset.** The CLI exposes every solver setting as a flag, generated from the parameter list at runtime: take any parameter documented in the cuOpt settings reference and replace underscores with hyphens (`time_limit` → `--time-limit`, `mip_relative_gap` → `--mip-relative-gap`). So if a parameter is documented, the flag exists; `--help` and the [solver-settings docs](https://docs.nvidia.com/cuopt/user-guide/latest/) are the sources of truth for names, meanings, and defaults.
 
-# MIP gap tolerance (stop when within X% of optimal)
-cuopt_cli problem.mps --mip-relative-tolerance 0.001
+A few options are CLI-specific (not solver parameters) and worth knowing because you wouldn't derive them from a parameter name:
 
-# MIP absolute tolerance
-cuopt_cli problem.mps --mip-absolute-tolerance 0.0001
+- `--params-file <file>` — supply many parameters from a `key = value` config file instead of repeating flags.
+- `--relaxation` — solve the continuous relaxation of a MILP (drop integrality).
+- `--initial-solution <file>` — warm-start from a solution file.
 
-# Presolve, iteration limit, method
-cuopt_cli problem.mps --presolve --iteration-limit 10000 --method 1
-```
+Run `cuopt_cli --help` for the complete, current set.
+
+## Authoring input files
+
+MPS, QPS, and LP are precise, externally-specified file formats. Don't hand-author them from memory or from a partial recollection of the layout — column-position rules, marker conventions, the quadratic-objective encoding, and sign/scaling conventions are easy to get subtly wrong, and a malformed file either fails to parse or **silently encodes a different model than intended**.
 
-## MPS format (required sections, in order)
+If you're building a model from data (rather than solving a file you already have), **define the problem through the cuOpt Python API or your preferred modeling interface — don't generate an MPS or LP file by hand to feed the CLI.** These interfaces take coefficients, bounds, and variable types as native data structures, so there's no text format to get wrong. The cuOpt Python API solves and returns the solution in the same program (see the `cuopt-numerical-optimization-api-python` skill); a separate modeling library (e.g. PuLP, Pyomo, JuMP) can export a valid file for the CLI or call its own solver. Reach for `cuopt_cli` when you *already* have a model file — e.g. a benchmark instance or a file exported by one of these tools.
 
-1. **NAME** — problem name
-2. **ROWS** — N (objective), L/G/E (constraints)
-3. **COLUMNS** — variable names, row names, coefficients
-4. **RHS** — right-hand side values
-5. **BOUNDS** (optional) — LO, UP, FX, BV, LI, UI
-6. **ENDATA**
+When a file is genuinely the right artifact, still generate it programmatically rather than by hand:
 
-Integer variables: use `'MARKER' 'INTORG'` before and `'MARKER' 'INTEND'` after the integer columns.
+- **From cuOpt's Python API** — build the model, then export it with `model.writeMPS("problem.mps")` (emits a valid MPS file, including the quadratic objective for QP) and solve with `cuopt_cli problem.mps`.
+- **From another modeling tool** — most LP/MILP modeling libraries and solvers can export standard MPS or LP files; pass the exported file straight to `cuopt_cli`.
 
-## QP via CLI (beta)
+If you must read or write these formats by hand anyway, work from the format's full specification (and the cuOpt repo docs at `docs/cuopt/source/cuopt-cli/` for cuOpt-specific conventions such as the quadratic-objective encoding) — not from an example alone.
 
-Quadratic objectives extend the standard MPS workflow — same `cuopt_cli` command, same options. Check `cuopt_cli --help` for QP-specific flags and the repo docs at `docs/cuopt/source/cuopt-cli/` for the quadratic-objective MPS format.
+Either way, **validate before trusting the result**: `cuopt_cli` logs `Read file ...` on a successful parse, and reports the variable/constraint counts and objective — sanity-check those against your intended model.
 
-**QP rules:**
-- **MINIMIZE only.** For maximization, negate the objective coefficients (and Q entries) in the MPS file.
-- **Continuous variables only** — do not mix integer markers with quadratic objectives.
+**QP notes (beta):** quadratic objectives are **MINIMIZE only** (for maximization, negate the objective, including the quadratic terms) and require **continuous variables only** (no integer variables mixed with a quadratic objective). Check `cuopt_cli --help` for QP-specific flags.
 
 ## Troubleshooting
 
-- **Failed to parse MPS** — Check ENDATA, section order (NAME, ROWS, COLUMNS, RHS, [BOUNDS], ENDATA), integer markers.
-- **Infeasible** — Check constraint directions (L/G/E) and RHS values.
+- **Parsing input file failed** — Confirm the extension matches the format (`.lp` vs `.mps`/`.qps`); an unrecognized extension is rejected before parsing. The parser error names the offending line — fix it against the format spec, or (more reliably) regenerate the file from a modeling tool rather than patching it by hand.
+- **Infeasible** — Re-check the model against your intended formulation: constraint directions, right-hand sides, and variable bounds.
 
 ## Examples
 

skills/cuopt-numerical-optimization-api-cli/skill-card.md

@@ -1,27 +1,33 @@
 ## Description: <br>
-LP, MILP, and QP (beta) with cuOpt — CLI only (MPS files, cuopt_cli). Use when the user is solving LP, MILP, or QP from MPS via command line. <br>
+LP, MILP, and QP (beta) with cuOpt — CLI only (MPS/LP/QPS files, cuopt_cli). Use when the user is solving LP, MILP, or QP from an MPS or LP file via command line. <br>
 
 This skill is ready for commercial/non-commercial use. <br>
 
 ## Owner
 NVIDIA <br>
 
 ### License/Terms of Use: <br>
-Apache 2.0 <br>
+Apache-2.0 <br>
 ## Use Case: <br>
-Developers and engineers solving LP, MILP, and QP optimization problems from MPS files via the cuopt_cli command-line interface. <br>
+Developers and engineers solving LP, MILP, and QP optimization problems from MPS or LP files via command line using NVIDIA cuOpt's GPU-accelerated solver. <br>
 
 ### Deployment Geography for Use: <br>
 Global <br>
 
+## Requirements / Dependencies: <br>
+**Requires API Key or External Credential:** [Not Specified] <br>
+**Credential Type(s):** [None identified] <br>
+
+Do not include secrets in prompts/logs/output; use least-privilege credentials; rotate keys as appropriate. <br>
+
 ## Known Risks and Mitigations: <br>
 Risk: Review before execution as proposals could introduce incorrect or misleading guidance into skills. <br>
 Mitigation: Review and scan skill before deployment. <br>
 
 ## Reference(s): <br>
-- [cuOpt User Guide](https://docs.nvidia.com/cuopt/user-guide/latest/introduction.html) <br>
-- [cuopt-examples](https://github.com/NVIDIA/cuopt-examples) <br>
-- [Sample MPS Assets](assets/README.md) <br>
+- [cuOpt User Guide](https://docs.nvidia.com/cuopt/user-guide/latest/) <br>
+- [cuOpt Examples](https://github.com/NVIDIA/cuopt-examples) <br>
+- [cuOpt Releases](https://github.com/NVIDIA/cuopt/releases) <br>
 
 
 ## Skill Output: <br>
@@ -31,13 +37,13 @@ Mitigation: Review and scan skill before deployment. <br>
 **Other Properties Related to Output:** [None] <br>
 
 ## Evaluation Agents Used: <br>
-- Claude Code (`claude-code`) <br>
-- Codex (`codex`) <br>
+- claude-code <br>
+- codex <br>
 
 
 
 ## Evaluation Tasks: <br>
-Evaluated against 1 evaluation task with 2 attempts per task via NVSkills-Eval (external profile, local environment). Pass threshold: 50%. <br>
+Evaluated against 1 evaluation task (positive skill-activation case) in NVSkills-Eval external profile. <br>
 
 ## Evaluation Metrics Used: <br>
 Reported benchmark dimensions: <br>
@@ -61,11 +67,11 @@ Underlying evaluation signals used in this run: <br>
 ## Evaluation Results: <br>
 | Dimension | Num | `claude-code` | `codex` |
 |---|---:|---:|---:|
-| Security | 2 | 100% (+0%) | 100% (+0%) |
-| Correctness | 2 | 100% (+0%) | 97% (+5%) |
-| Discoverability | 2 | 100% (+0%) | 84% (+5%) |
-| Effectiveness | 2 | 78% (+2%) | 76% (+4%) |
-| Efficiency | 2 | 93% (-0%) | 78% (-0%) |
+| Security | 1 | 100% (+0%) | 100% (+0%) |
+| Correctness | 1 | 30% (+0%) | 77% (+47%) |
+| Discoverability | 1 | 0% (+0%) | 84% (+84%) |
+| Effectiveness | 1 | 39% (-4%) | 41% (+4%) |
+| Efficiency | 1 | 27% (-0%) | 77% (+49%) |
 
 ## Skill Version(s): <br>
 26.08.00 (source: frontmatter) <br>