# Thalarch Spec

> Turns broad feature or architecture requests into an executable acceptance contract. Use for multi-file features, refactors, migrations, architecture work, ambiguous behavior, or whenever implementation could succeed technically while missing the user's actual intent.

- Skill: `luc4n3x/thalarch-spec` (Agent Skill)
- Install (CLI): `npx skillmds@latest add luc4n3x/thalarch-spec`
- Raw SKILL.md: https://api.skillmd.com/api/skills/luc4n3x/thalarch-spec/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: LUC4N3X (https://skillmd.com/u/luc4n3x)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/luc4n3x/thalarch-spec

---


# Thalarch Spec

Treat requirements as testable constraints, not prose.

## Build the acceptance matrix

For each user requirement record:
- ID;
- observable behavior;
- scope;
- proof;
- non-goal;
- dependency;
- ambiguity/ruling.

Separate:
- functional behavior;
- compatibility;
- error behavior;
- performance constraints;
- UI/UX constraints;
- external-action constraints.

## Clarify only material ambiguity

Ask only if different interpretations would materially change the outcome and no
safe reversible default exists.

Otherwise record a ruling with:
`Ruling — reason — cost if wrong`.

## Cross-artifact consistency

Before implementation, check:
- no requirement is missing a planned task;
- no planned task exists without a requirement or justified engineering necessity;
- tests actually prove the requirement they claim to cover;
- two tasks do not define conflicting interfaces.

## Output

The specification should be compact enough to hand directly to a cold verifier.