Canonical: https://botbento.com/usecases/score-feature-specs-before-coding-agents-start/
Format: Markdown representation of the public HTML page.

# Score feature specs before coding agents start

[Coding](https://botbento.com/usecases/coding/) · Intermediate

Catch missing scope, edge cases and tests in a written spec

A bot reads a feature spec and grades it against a fixed checklist before anyone hands it to a coding agent. It marks unclear goals, undefined data shapes, unstated edge cases and missing acceptance tests, then turns each gap into a short clarifying question so the spec can be tightened while changes are still cheap.

## Quick answers

**What does it do?** Catch missing scope, edge cases and tests in a written spec. Result: Scored report with open questions.

**What does it need?** Tools: Docs, Notes. Intermediate: some setup. The prompt includes an approval step.

**How do I run it?** When something happens: Spec document updated. Copy the prompt below into a bot in BotBento.

## How it works

1. When something happens: Spec document updated
2. Then: Score spec against checklist
3. Result: Scored report with open questions

## Prompt to start with

```text
Read the spec document the user points you to. Score it against a fixed checklist: goal, scope boundaries, data shapes, error and edge cases, acceptance tests, and rollout notes. For every gap, cite the section heading and write one clarifying question. Give a score per section plus an overall pass or revise verdict, and keep drafted rewrites in a separate note rather than editing the original. Do not send, post or delete anything without the user's approval.
```

## Tools it uses

Docs, Notes

## Works with

BotBento

## Tags

specs, checklist, quality, review

## Source

[Epic framework for spec-driven development](https://news.ycombinator.com/item?id=50392336) — Hacker News · checked 2026-11-13

Summarized in our own words. Any results described there are the authors’ own, not measured by BotBento.

Verified 2026-11-13: passed every automated gate and a second-model review (kimi-code/k3.1-pro). Automated, not human review. [How verification works](https://botbento.com/usecases/verification/)

By BotBento Editorial · Added 2026-11-13 · Updated 2026-11-13 · [Suggest a correction](https://botbento.com/contact/)

## Related use cases

- [Draft executable feature specs before any code is generated](https://botbento.com/usecases/draft-executable-feature-specs-before-any-code/) — Turn a rough feature request into a reviewable spec file
- [Spec-driven feature implementation from task queues](https://botbento.com/usecases/spec-driven-feature-implementation-from-task-queues/) — Turn plain-text feature specs and task tracker issues into draft branches and pull requests.
- [Review agent-written pull requests against quality gates](https://botbento.com/usecases/review-agent-written-pull-requests-against-quality/) — Check machine-generated diffs for tests, style and structural fit
