Validate and ablate readable NLAH.md harness policy without executing ambiguous prose
Scanned 9/3/2026
Install to Claude Code
Skills Directorynpx -y skills add jmagly/aiwg --skill natural-language-harness --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Natural Language Harness?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/jmagly-natural-language-harness)More formats (shields.io, HTML) on the badges page.
---
namespace: aiwg
name: natural-language-harness
platforms: [all]
description: Validate and ablate readable NLAH.md harness policy without executing ambiguous prose
---
# Natural-Language Harness
Use an NLAH document to review harness policy and its deterministic mappings.
1. Write the seven required sections.
2. Mark executable clauses as `MUST [id]`.
3. Map every ID to a validator, script, agent, flow, or manual gate.
4. Run `aiwg harness validate NLAH.md` before planning.
5. Use `aiwg harness ablate NLAH.md --remove <module>` to compare modules.
Never execute prose directly. `SHOULD` and `MAY` clauses remain ambiguous and
must be reviewed or promoted to explicit mappings before execution.
@implements #2043
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!
