sddl-eval-level-1

Runs deterministic structural integrity checks against an installed sdd-lite workspace.

Updated Apr 2, 2026

Search Tags:#testing #evaluation #workspace #deterministic #integrity check #structural validation #sdd-lite

Ready to run

npx skills add https://github.com/nico0695/ai-tools --skill sddl-eval-level-1-nico0695

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: sddl-eval-level-1
Source: https://github.com/nico0695/ai-tools/tree/main/sdd/sdd-lite-evals/skills/sddl-eval-level-1
Command: npx skills add https://github.com/nico0695/ai-tools --skill sddl-eval-level-1-nico0695

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It catches structural inconsistencies in an sdd-lite installation—broken wrappers, schemas, routes, skill files, bootstrap paths, or legacy runtime references—before any model-backed evaluation runs, preventing wasted effort on invalid setups. ## Core Features & Use Cases - Zero-Token Structural Gate: Executes a deterministic level-1 check via the sddl_eval.py harness without consuming model tokens. - Workspace Resolution: Recovers the workspace id from workspaces/*/project.yaml instead of requiring manual input. - Verdict Reporting: Reads the generated report.md and surfaces verdict, failures, warnings, and direct evidence paths. - Use Case: After modifying sdd-lite wrappers or schemas, run this gate to confirm the installation is structurally sound before proceeding to Levels 2 and 3 evaluations. ## Quick Start Run the level 1 structural integrity check on my sdd-lite workspace and show me the verdict with any failures and evidence paths.

Frequently Asked Questions about sddl-eval-level-1

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run structural integrity checks on sdd-lite?▼

Run python3 <harness>/scripts/sddl_eval.py level1 --workspace <id> after resolving the harness via sddl-eval-init. The workspace id can be recovered from workspaces/*/project.yaml, and results appear in the generated report.md.

What does the sdd-lite level 1 evaluation check?▼

It performs deterministic structural checks on wrappers, schemas, routes, skill files, bootstrap paths, and legacy runtime references. It runs without model tokens and gates all higher evaluation levels.

Can I run level 2 or 3 evaluations after a level 1 failure?▼

No. BLOCKED and FAIL verdicts prevent Levels 2 and 3 from running. A WARN verdict permits them only when the warning cannot invalidate the intended test case.

Why is my sdd-lite workspace id not found?▼

The workspace id is recovered from workspaces/*/project.yaml files. If none exists, the workspace may not be initialized; verify the project.yaml file exists under a workspaces subdirectory before asking for manual input.

Should I read raw inputs or the report after level 1 evaluation?▼

Read the generated report.md rather than every raw input. It contains the verdict, failures, warnings, and direct evidence paths needed to assess structural integrity.