9 views
This guide focuses on prompt change management for production systems, with practical trade-offs for reliability, security, and cost.
Hot-editing the system prompt in a Lambda env var is how you ship an incident without a diff. Prompts are code: version, review, eval, pin.
⚡ TL;DR: Store prompts in Bedrock Prompt Management or git. Pin versions in the gateway. Promote with CI eval gates. No console edits in prod.
prompt change management: production guidance
text
PR (prompt + eval delta) → CI eval → stage pin → prod pin → monitor
python
# ✅ Gateway resolves pin, not "latest"
def resolve_prompt(prompt_id: str, env: str) -> tuple[str, str]:
pin = registry.get_pin(prompt_id, env) # e.g. "v12"
body = registry.fetch(prompt_id, pin)
return pin, body
| Rule | Why |
|---|---|
| Diff in PR | Reviewable |
| Eval required | Catch regressions |
| Pin in gateway | Determinism |
| Rollback = repin | Minutes, not archaeology |
PROMPT=$(curl raw.githubusercontent.com/.../main/...) at runtime.
Failure modes
Two gateways pinning different versions after a partial deploy. Pin config in a single store with atomic swaps. “Emergency” console edits without a follow-up PR — detect drift jobs that diff live vs git.Closing checklist
- [ ] Prompt registry with versions
- [ ] Gateway pins per env
- [ ] CI eval on prompt PRs
- [ ] Console write access removed in prod
- [ ] Rollback runbook = previous pin
Series navigation
Day 93: SLOs for AI Features · Day 95: Incident Response When the Agent Goes Wrong Last updated September 11, 2026 Related: Bedrock Prompt Management: Versioned System Prompts for Production; Day 61: Prompt Injection in Jira, Email, and RAG; Day 72: Prompt Caching Pitfalls.Watch: 100 Days of AI on YouTube
Short videos from CheatCoders, one AI engineering topic per day.
Most viewed
- Python Decorators Explained: From Simple Wrappers to Production Patterns
- Claude Code vs Cursor vs GitHub Copilot: Choose by Workflow, and Set Up One Repo That Works With All Three
- AI Agent Frameworks in 2025: LangGraph vs CrewAI vs AutoGen vs Raw API
- Python String Methods: Every str Method With Real Production Examples
- Css Animation Performance: Performance-Optimized Motion for Modern UIs
Newly added
- AWS Systems Manager Parameter Store: Hierarchical Runtime Config for Coding Agents
- Amazon Cognito Identity Pools: Federate Coding-Agent Tool Callers Without Long-Lived Keys
- Amazon Athena for Coding Agents: Query Tool Audit Logs and Eval Traces on S3 Without a Warehouse
- AWS Batch for Coding Agents: Parallel Overnight PR Reviews and Long-Running Jobs Without Lambda Timeouts
- Amazon Bedrock Prompt Management: Version and A/B Test Coding-Agent System Prompts
Deep-dive PDF
Get the expanded guide for this post — extra diagrams-style checklists, failure modes, and a production walkthrough. Free when you subscribe to CheatCoders.
Already subscribed? or open the subscribe page.
Discover more from CheatCoders
Subscribe to get the latest posts sent to your email.
