Agent skill · momentic-ai
momentic-spec
Improve code correctness using Momentic specs in the feature development process
What it needs
About 6k tokens when loaded.
What this skill does
Momentic spec-driven development Write the expected product behavior as Momentic tests before implementing it. The tests define the contract; do not make the tests fit the implementation. Non-negotiable sequence 1. Read repository guidance, momentic.config.yaml, and nearby tests and modules. 2. Identify and report the exact Momentic tests expected to be affected. 3. Create or modify those tests before changing product code. 4. Review the test diff as a product specification. 5. Only then implement the product change. 6. Use fast code-level checks while implementing; do not run Momentic after every edit or commit. 7. At a durable end-to-end checkpoint, prepare the required test state, enable the affected tests that should now pass, and ask whether to run the smallest affected set. Do not begin product implementation until the intended test specifications are on disk. If the user changes the requirement, update the spec first again. Choose the affected tests If the user names specific existing tests, start with those tests and keep the work focused on them. Expand the set only when repository evidence shows another test or shared module is directly affected. For new user-visible functionality, create the smallest new test or tests that describe the new user journey. For changed functionality, update the existing tests that describe the old behavior instead of adding duplicate coverage. Trace shared components and modules far enough to find real behavioral impact. Do not infer the affected set from filenames alone. If no Momentic test is affected, state why before making product changes. Do not make cosmetic test edits merely to produce a test diff. Before editing product code, report each affected test path and whether it will be added, changed, or left unchanged. Write the product specification first The pre-implementation test should describe the requested outcome, even when the current product cannot satisfy it yet. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills momentic-ai/momentic-spec