Overview

Treat Blink like a new teammate who can research quickly, draft options, and execute once the plan is clear. Your job is to set direction, provide context, and keep scope small. Start broad, then get specific, and iterate in short cycles.

Before You Start

  • Confirm GitHub is connected and Blink can access the target repositories
  • Collect context up front: problem statement, acceptance criteria, constraints, code links
  • Find small, well-defined tasks over broad multi-epic prompts

Workflow

  1. Frame the problem and outcome
  2. Ask Blink for 2–3 approaches with trade-offs
  3. Pick an approach and define acceptance criteria
  4. Break into small tasks with PR-sized milestones
  5. Identify risks, unknowns, and checkpoints
  6. Lock the plan and begin executing task 1

Prompt Patterns That Work

  • Break complex asks into smaller steps
  • Avoid ambiguity (reference exact files, functions, or URLs)
  • Iterate: refine after each response

Effective Prompts

You are planning work for a small feature.
Context:
- Repo: <public link or description>
- Area: notifications service (Go), stores messages in Postgres
- Problem: daily summary email sometimes duplicates entries
- Constraints: keep DB schema stable
- Success: no duplicates; add an integration test; draft PR

Task: Propose 2–3 approaches with trade-offs (risk, complexity, testability). Then propose a step-by-step plan for the lowest-risk path with clear acceptance criteria and a first milestone suitable for a small PR.
We’re evaluating background job frameworks.
Context:
- Framework: Node.js 20
- Current stack: Express, Prisma, Postgres
- Needs: retries, delayed jobs, metrics, horizontal scaling

Task: Compare BullMQ vs. Bree vs. Temporal (self-hosted). Provide: pros/cons, code snippets for a minimal worker, migration risk, and the safest recommendation for our team skill level. End with a suggested proof-of-concept plan (1–2 days).

Anti‑Patterns and Recovery

  • Vague asks (“make it better”): restate the problem, add a clear outcome
  • One giant prompt: split into plan → milestone 1 → iterate
  • Missing constraints: state performance, security, compliance, and team preferences
  • Clear menu of approaches with trade-offs
  • A small, testable milestone plan that’s easy to review
  • Follow-up questions if requirements are unclear