Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add rules/technickai/claude_telemetry/prompt-engineeringgit clone --depth 1 https://github.com/TechNickAI/claude_telemetryWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00015 | $0.01553 |
| Opus 5 | $0.00008 | $0.00776 |
| Sonnet 5 | $0.00003 | $0.00311 |
| Haiku 4.5 | $0.00002 | $0.00155 |
Grade A, and why
prompt-engineering scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 220 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Prompt Engineering Best Practices
Understanding LLM Token Prediction
How LLMs Actually Process Prompts
Understanding the mechanics of LLM processing is crucial for effective prompt design:
Sequential Token Prediction
- LLMs read left to right, predicting each token based on everything before it
- Early tokens shape all that follows—creating a "first impression" that persists
- Each token's prediction is influenced by ALL previous tokens, creating cascading effects
Attention Mechanisms
- Earlier tokens receive more attention passes during processing
- The model repeatedly references early context when interpreting later content
- This creates a foundation effect where initial framing heavily influences all subsequent reasoning
Context Window Effects
- Primacy: Information at the beginning is strongly encoded and influences everything
- Recency: Information near the end is fresh in "working memory" for decisions
- Middle Fade: Information in the middle can get "lost" if not properly structured
Priming & Anchoring
- Early statements act as anchors that bias interpretation of everything else
- The model's "persona" crystallizes early and remains consistent throughout
- Initial framing determines the lens through which all data is viewed
Implications for Prompt Design
SYSTEM PROMPT (Static Foundation) - the agent's DNA:
- Identity First: Who the agent IS fundamentally shapes HOW it thinks
- Core Principles: Unwavering rules or beliefs that guide decisions
- Operational Framework: Methodology for the task
- Capabilities & Constraints: What the agent can and cannot do
USER PROMPT (Dynamic Context) - real-time inputs:
- Current Context: Immediate situation or environment
- Specific Data: The information to process
- Task Request: Clear ask with expected output format
Pattern Reinforcement Through Examples
LLMs learn patterns from what you show them, regardless of labels.
How LLMs Process Examples
When you provide code examples, LLMs encode those patterns whether you label them positively or negatively:
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 220 lines · 15 tokens per session scan A f8f9690f716c
prompt-engineering is a cursor rule published in the GitHub repository TechNickAI/claude_telemetry (30 stars, last pushed 10mo ago), licensed MIT. It adds 15 tokens to every session and 1,553 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other cursor rules, from other repositories
orchestrator-sub
Rules for sub-agents spawned to work on specific tasks.
testing-standards-typescript
When writing tests for typescript.
external-apis
When calling external APIs.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
coolify-ai-docs
Master reference to all Coolify AI documentation in .ai/ directory.