Nm Leyline Evaluation Framework
v1.0.0Patterns for building evaluation and scoring systems, quality gates, rubrics, and decision frameworks. Use for any scored assessment
⭐ 0· 44·1 current·1 all-time
by@athola
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name, description, and included modules (scoring-patterns, decision-thresholds) match: the skill is an authoring/reference framework for scoring and thresholds and does not request unrelated capabilities or credentials.
Instruction Scope
SKILL.md contains only patterns, examples, and pseudocode; it does not instruct the agent to read files, call external endpoints, or access secrets. Minor inconsistency: many examples include verification steps like 'Run the command with --help' or 'Run pytest -v', but no concrete command, tests, or dependencies are provided in the package—this appears to be template text for consumers rather than active instructions to access system resources.
Install Mechanism
No install spec or code files are present (instruction-only). Nothing is downloaded or written to disk by the skill itself.
Credentials
The skill declares no required environment variables, credentials, or config paths. The content contains no references to hidden tokens or unrelated service credentials.
Persistence & Privilege
Skill is not always-on and is user-invocable. It does not request elevated persistence or modify other skills or system settings.
Assessment
This skill is documentation-only: it provides templates and code examples for building scoring/threshold systems and does not require credentials or install anything. Before using it in an automated agent: (1) review and adapt the example commands/tests—the SKILL.md includes generic 'run --help' and 'pytest -v' checks but no bundled binaries or tests; (2) if you intend to automate decisions based on these thresholds, validate them with historical data and add veto/safety checks to avoid unwanted automated actions; (3) the agent may call this skill autonomously (normal default), so ensure decision rules and escalation paths are explicit and safe for your environment.Like a lobster shell, security has layers — review code before you run it.
latestvk97b0qg015yckcm7fy47tn6tks84rtcf
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🦞 Clawdis