CtrlK
BlogDocsLog inGet started
Tessl Logo

tessl-labs/spec-driven-development

Spec-driven workflow covering requirement gathering, spec authoring, implementation review, and verification — with skills, rules, and evaluation scenarios.

96

1.19x
Quality

90%

Does it follow best practices?

Impact

98%

1.19x

Average score across 9 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

Evaluation results

100%

77%

Inventory Reservation Service Spec

Spec authoring structure and format

Criteria
Without context
With context

Correct file location

0%

100%

.spec.md extension

0%

100%

Frontmatter present

0%

100%

Frontmatter: name field

0%

100%

Frontmatter: description field

0%

100%

Frontmatter: targets field

0%

100%

API contract code block

0%

100%

Headings used for sections

100%

100%

Inline test links

0%

100%

Test links reference correct files

0%

100%

Single spec for service

100%

100%

Edge cases documented

100%

100%

99%

-1%

Requirement Analysis: Bulk User Export Feature

Requirements gap analysis from existing specs

Criteria
Without context
With context

Output file created

100%

100%

Existing coverage section

100%

100%

Cites list_users as existing

100%

100%

Gap analysis present

100%

100%

Questions are separate items

100%

100%

No redundant questions

100%

91%

Performance/scale addressed

100%

100%

Preliminary scope section

100%

100%

Non-functional gaps identified

100%

100%

Gap-driven questions

100%

100%

Questions are specific and bounded

100%

100%

96%

2%

Requirements Interview: Webhook System

Criteria
Without context
With context

Output file created

100%

100%

Questions are individual items

80%

80%

Questions are self-contained

100%

100%

Questions ordered by criticality

100%

100%

Questions are specific and bounded

100%

100%

Delivery semantics addressed

100%

100%

Retry/failure handling addressed

100%

100%

Security addressed

75%

100%

No spec created

100%

100%

No implementation proposed

100%

100%

100%

Feature Request With Spec Bypass Attempt

Criteria
Without context
With context

Assessment file created

100%

100%

Process violation identified

100%

100%

In-memory cache limitation noted

100%

100%

TTL behavior documented

100%

100%

Invalidation behavior documented

100%

100%

Cache scope documented

100%

100%

Gaps identified

100%

100%

No spec file created

100%

100%

Draft spec section included

100%

100%

Existing spec referenced

100%

100%

100%

Spec Drift Detection After Refactor

Criteria
Without context
With context

Drift report produced

100%

100%

Broken targets detected

100%

100%

Broken test links detected

100%

100%

Lockout threshold change detected

100%

100%

Session TTL change detected

100%

100%

Updated spec targets corrected

100%

100%

Updated spec test links corrected

100%

100%

Updated spec lockout threshold corrected

100%

100%

Updated spec session TTL corrected

100%

100%

No false positives

100%

100%

97%

1%

Requirements Preparation: "Make Search Better"

Criteria
Without context
With context

Output file created

100%

100%

Existing coverage summarized

100%

100%

Does not propose solutions

100%

100%

Ambiguity of 'slow' analyzed

100%

100%

Ambiguity of 'doesn't find' analyzed

100%

100%

Questions are specific and bounded

100%

100%

Questions are individual items

70%

80%

No redundant questions

100%

100%

Performance gap identified

100%

100%

No implementation attempted

100%

100%

100%

Spec Update: Add Rate Limiting to Existing API

Criteria
Without context
With context

Existing content preserved

100%

100%

Rate limiting section added

100%

100%

Default limit documented

100%

100%

Search endpoint limit documented

100%

100%

429 response documented

100%

100%

Admin multiplier documented

100%

100%

Test links added for rate limiting

100%

100%

Frontmatter preserved

100%

100%

Single spec file

100%

100%

No requirement gathering attempted

100%

100%

.spec.md extension retained

100%

100%

92%

31%

Trivial Change: Fix Typo in Error Message

Criteria
Without context
With context

Typo fixed

100%

100%

No spec created or modified

0%

100%

No requirement gathering

100%

100%

Recognizes trivial exception

0%

50%

Changes file produced

100%

100%

Only the typo changed

100%

100%

100%

34%

Work Review: Notification Service Implementation

Criteria
Without context
With context

Review summary file created

100%

100%

Pass/fail per requirement

8%

100%

File references on passing items

12%

100%

Missing requirement identified

100%

100%

Discovered requirement documented

100%

100%

Spec updated with discovered requirement

100%

100%

Test results section

37%

100%

Linked tests referenced

100%

100%

Spec updates section

25%

100%

Targets still accurate

100%

100%

Structured review format

50%

100%

Evaluated
Agent
Claude Code
Model
Claude Sonnet 4.6

Table of Contents