CtrlK
BlogDocsLog inGet started
Tessl Logo

spec-driven-devlopment/spec-as-source

Spec-driven development on OpenSpec, with mechanical spec-as-source enforcement: a custom 'spec-as-source' OpenSpec schema adds file-ownership (targets) and test-verification ([@test]) metadata to every capability spec, three scripts (link check, ownership check, manifest build) keep code and specs from drifting apart, plus requirement-gathering, spec-writer, work-review, and a session-handoff skill with a proactive context-warning hook.

73

Quality

92%

Does it follow best practices?

Impact

No eval scenarios have been run

SecuritybySnyk

Advisory

Suggest reviewing before use

Overview
Quality
Evals
Security
Files

REFINED_PROMPT.mdskills/prompt-loop/templates/

Refined Prompt — Output Contract

The final artifact of a prompt-loop session, produced only after the loop reaches its stop criterion. It contains exactly the sections the SDD workflow consumes downstream, so it feeds openspec-propose or openspec-explore directly, with no re-asking of locked decisions.

Every section is filled from the session document or marked n/a — never invented, assumed, or filled with plausible defaults. Present this document to the user for explicit confirmation before any downstream step runs.

Refined prompt

The final prompt revision, verbatim from the last round of the session document. All locked decisions appear in it unchanged.

Scope

In scope

Out of scope

Core behavior

What the deliverable must do, stated as observable behavior — the essence that openspec-propose will turn into requirements.

Edge cases

Boundary conditions, failure modes, and unusual inputs the user confirmed or that the prompt states. Unknowns are n/a, not guesses.

Constraints

Technical, stylistic, or process constraints the work must respect (stack, conventions, compatibility, performance, deadlines).

Capabilities touched

Existing or new capabilities affected, in kebab-case — the candidate openspec/specs/<capability>/ names downstream.

  • (new | modified)

Verification expectations

How the user will know the work is done: tests to exist, behaviors to demonstrate, checks to pass.

Final rubric score

The last scoring round, copied from the session document. Every score above 0 keeps its verbatim evidence quote.

DimensionScore (0–2)Evidence (verbatim quote)
Objective clarity
Scope boundaries
Verifiability of outcomes
Completeness of constraints and edge cases
Absence of invented or assumed facts

Total: <0–10>

Lock register

The complete ## Locked register from the session document, copied verbatim — including superseded entries kept struck-through for audit.

  • L1 (round ): "<user's answer, verbatim>"

skills

README.md

tile.json