CtrlK
BlogDocsLog inGet started
Tessl Logo

spec-miner

Reverse-engineering specialist that extracts specifications from existing codebases. Use when working with legacy or undocumented systems, inherited projects, or old codebases with no documentation. Invoke to map code dependencies, generate API documentation from source, identify undocumented business logic, figure out what code does, or create architecture documentation from implementation. Trigger phrases: reverse engineer, old codebase, no docs, no documentation, figure out how this works, inherited project, legacy analysis, code archaeology, undocumented features.

97

1.21x
Quality

100%

Does it follow best practices?

Impact

95%

1.21x

Average score across 6 eval scenarios

SecuritybySnyk

Risky

Do not use without reviewing

SKILL.md
Quality
Evals
Security

Evaluation results

100%

47%

Reverse-Engineer a Legacy Order Processing Service

Specification output format and required sections

Criteria
Without context
With context

Correct output path

0%

100%

Technology stack section

87%

100%

Module structure section

37%

100%

EARS format requirements

0%

100%

Code locations cited

100%

100%

Non-functional section

100%

100%

Acceptance criteria section

0%

100%

Uncertainties section

100%

100%

Recommendations section

0%

100%

No assumptions without evidence

100%

100%

Facts vs inferences distinguished

62%

100%

100%

Document a Payment Gateway Integration Service

Dual-hat analysis and security/error handling coverage

Criteria
Without context
With context

Arch Hat perspective

100%

100%

QA Hat perspective

100%

100%

Security patterns documented

100%

100%

Error handling documented

100%

100%

Error response format noted

100%

100%

Auth flow traced

100%

100%

Input validation behavior

100%

100%

Code evidence for security

100%

100%

Code evidence for error handling

100%

100%

No skipped security section

100%

100%

No skipped error handling

100%

100%

92%

-2%

Create a Handoff Specification for an Acquired SaaS Service

Core workflow, acceptance criteria, data flow, and recommendations

Criteria
Without context
With context

Scope step documented

100%

50%

Explore step with tools

40%

60%

Trace step documented

100%

100%

Flag step present

100%

100%

Given/When/Then criteria

100%

100%

Data flow diagram or description

100%

100%

Recommendations provided

100%

100%

API surface documented

100%

100%

Data models documented

100%

100%

Background jobs or async checked

100%

100%

External dependencies listed

100%

100%

Test coverage noted

100%

100%

78%

50%

Reverse-Engineer the Order Management Service

EARS format type variety

Criteria
Without context
With context

Event-Driven EARS used

0%

100%

State-Driven EARS used

0%

0%

Conditional EARS used

0%

100%

Optional EARS used

0%

100%

Ubiquitous EARS used

0%

0%

Code locations cited

0%

100%

Correct output path

100%

100%

State transitions documented

100%

100%

Uncertainties section present

100%

100%

100%

5%

Document the Auth Service Before the On-Call Rotation

Test file analysis and exploration thoroughness

Criteria
Without context
With context

Password min-length documented

100%

100%

Uppercase requirement documented

100%

100%

Numeric requirement documented

100%

100%

Lock threshold documented

100%

100%

Lock duration documented

100%

100%

Token TTLs documented

100%

100%

Code locations cited

100%

100%

Exploration log written

100%

100%

Facts vs inferences labelled

37%

100%

Uncertainties section present

100%

100%

Correct output path

100%

100%

100%

Spec the Payment Worker Before the Cloud Migration

External integrations and configuration documentation

Criteria
Without context
With context

Stripe integration documented

100%

100%

SendGrid integration documented

100%

100%

Required env vars documented

100%

100%

Optional env vars documented

100%

100%

Payment batch schedule documented

100%

100%

Daily report schedule documented

100%

100%

Timeout behavior documented

100%

100%

Code locations cited

100%

100%

Correct output path

100%

100%

Uncertainties section present

100%

100%

Non-functional observations present

100%

100%

Repository
jeffallan/claude-skills
Evaluated
Agent
Claude Code
Model
Claude Sonnet 4.6

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.