Guides

Let Nightshift choose the highest-value work

Use Automatic Hunt to inspect a project, rank applicable work, and fill a bounded shift without inventing a backlog.

Answers: automatically find the best coding work in a project · Updated

SHORT ANSWER

Give Nightshift a time budget and choose review first or run directly. It inspects the actual work target, selects only compatible catalog contracts, ranks them by evidence, user value, risk, and time, removes overlap, and keeps useful remaining time productive.

01

Choose how much authority to give

Review first completes a read-only survey and waits with an ordered proposal. Run directly treats the declared time and safety boundary as authority to select and execute reversible work immediately.

  • Review first — inspect and rank, then wait.
  • Run directly — inspect, compose one contract, arm, and work.
  • Both modes require a fixed time budget for open-ended work.
02

Selection follows the project

Automatic Hunt reads the work target rather than assuming every catalog entry applies. It skips repository-only work in artifact folders, requires existing tools where a shift depends on them, and avoids duplicate findings.

  • Evidence before priority.
  • Finite work before open-ended continuation.
  • No manufactured tasks merely to consume time.
03

External authority stays separate

Direct mode can make strong reversible decisions inside the isolated work boundary and records significant choices. It does not silently gain authority to merge, deploy, publish, spend money, expose secrets, or alter legal and licensing policy.

Run Nightshift Hunt in Automatic mode for four hours, review first. Rank applicable work and show the proposed contract without starting it.
TRY IT

Start with a bounded prompt

This prompt names the outcome and preserves Nightshift’s review boundary. Paste it into the supported coding host from the project you want to change.

Run Nightshift Hunt in Automatic mode for four hours, review first. Rank only applicable work by evidence, user value, risk, and available time.
BOUNDARIES

What this workflow does not claim

  • Automatic selection skips catalog entries the work target cannot support and must not manufacture work to fill the clock.
  • Run directly is execution authority inside the declared boundary, not authority to merge, deploy, publish, spend money, or expose secrets.
SOURCES

Evidence and sources

These links support the released behavior, public outcomes, or problem language described on this page.

Hunt skillGuided and automatic selection, review-first, run-direct, and clock rules.Open evidence ↗Released ready-shift catalogEvery ready shift ships as a bounded Markdown contract in the plugin.Open evidence ↗Workflow contractThe shipped one-item loop, decision boundary, gates, recovery, and clock-out behavior.Open evidence ↗
RELATED QUESTIONS

Continue from the question you have

let a coding agent choose useful workRead the answer →review coding-agent findings before changesRead the answer →run product improvements for a fixed timeRead the answer →