Agentic Sdlc URL Shortener — Documentation skill for Claude Code

Agentic Software Engineering System: URL Shortener

An intentionally small, local prototype for the AI Engineering Assessment. It contains a working URL-shortener API and an offline SDLC orchestrator that turns a requirement into a reviewable plan, design, risk analysis, implementation proposal, validation summary, and release-readiness decision.

The orchestration provider is deterministic. It demonstrates workflow control, not live LLM reasoning or automatic code generation. A narrow provider interface makes later model integration possible without making this submission depend on a key, network, or paid service.

Requirements

• Python 3.11 or newer
• No runtime third-party packages

Run the tests

From this project directory, in PowerShell:

py -3 -m unittest discover -s tests -v

Run the orchestration scenarios

Run each scenario and inspect the JSON report:

py -3 -m agentic_shortener run --scenario greenfield
py -3 -m agentic_shortener run --scenario brownfield
py -3 -m agentic_shortener run --scenario ambiguous

Sample outputs for all three paths are included under `examples\reports\`.

The greenfield and brownfield scenarios reach the release gate and pause until approved. To record human approval of the release-readiness action:

py -3 -m agentic_shortener run --scenario greenfield --approve publish_release --output artifacts\greenfield-approved.json

The approval is recorded in the workflow report; the prototype does not publish or deploy anything. The ambiguous scenario stops before implementation until its open questions are answered. It can be rerun with an edited scenario file:

py -3 -m agentic_shortener run --scenario-file path\to\my-scenario.json

Every run fingerprints the requirement, creates a new run ID, and recomputes dependent stages rather than reusing stale stage output.

Run the URL-shortener API

Start the local service:

py -3 -m agentic_s