Free field guide · Merit by Moonlux

Legal AI Work Starter Kit

Find the work. Prepare clearly. Finish with fewer preventable errors.

The honest boundary: Moonlux tracks public openings and helps lawyers prepare for the workflow. A listing is not a promise of work. Every platform and project has its own eligibility, tool and confidentiality rules.
01

Find a fitting opening

  • Confirm jurisdiction, active-admission and location requirements from the original listing.
  • Distinguish an active project from a general talent-network application.
  • Compare the work requested—not only the headline hourly rate.
See Moonlux’s current observed listings →
02

Prepare evidence of expertise

  • Use accurate practice-area and work-product language that a matching system can understand.
  • Prepare three non-confidential examples: the problem, your role, your reasoning and the outcome.
  • Expect an assessment to test judgment, authority, clarity and instruction-following—not AI buzzwords.
Learn Build, Break, Answer and Grade →
03

Map the task before doing the task

  • Identify the exact final deliverables, sources, current instructions and permitted tools.
  • For multi-file work, record frozen facts, dependencies, named artifacts and approved exceptions.
  • Remember: attached does not mean read. Verify that every required input actually entered your process.
Use the world-building phase planner →
04

Run the final pass

  • Separate final deliverables from sources, notes and old versions.
  • Trace quotations, citations and claimed failures back to the source material.
  • Check cross-file facts, dates, names, metadata and every test that did—or did not—run.
  • Keep the lawyer in the loop: a checker can surface risk; it cannot decide whether the legal analysis is right.
Open the full checklist and private sample →

One clean loop

Scope → gather → reason → produce → verify → submit.

If a repair changes an upstream fact, rerun the downstream checks. Do not patch one file and assume the world still holds.