Menu
Manage AI agent context effectively — what to include, what to exclude, compression strategies, and context hierarchy for optimal performance.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Build USSD applications for African markets — menu design, session management, multi-operator integration, and user experience patterns for feature phones.
Generate AI Act Article 50 compliant disclosure notices, metadata tags, and content marking for AI-generated text, images, audio, and video.
Analyze contracts to identify key clauses, risks, obligations, deadlines, and parties. Multi-language support with focus on French and EU commercial law.
Detect and classify technical debt in AI-generated code — patterns specific to LLM outputs, shallow implementations, missing edge cases, and accumulation signals.
Navigate French notarial acts — fee calculation (emoluments), succession, donation, SCI creation, property transfer, and regulated fee schedules.
Structure a job search for the French/European market — CV format europass, cover letter VOUS-MOI-NOUS, job evaluation scoring, interview prep STAR, salary negotiation with French conventions.
| name | context-engineering |
| description | Manage AI agent context effectively — what to include, what to exclude, compression strategies, and context hierarchy for optimal performance. |
| version | 1.0.0 |
| last-updated | 2026-04-17 |
| model_tested | claude-sonnet-4-6 |
| category | meta |
| platforms | ["claude-code","codex","gemini-cli","cursor","copilot","windsurf","cline"] |
| language | en |
| geo_relevance | ["global"] |
| priority | high |
| dependencies | {"mcp":[],"skills":[],"apis":[],"data":[]} |
| update_sources | [{"url":"https://www.augmentcode.com/guides/how-to-build-agents-md","check_frequency":"quarterly","last_checked":"2026-04-17"}] |
| license | MIT |
Based on ETH Zurich research: overly detailed instructions reduce task success by 3%, increase token cost by 20%, and add 2-4 reasoning steps.
Most persistent → most transient:
| Level | Content | Persistence | Example |
|---|---|---|---|
| 1. Rules | Project-wide standards | Always loaded | CLAUDE.md, AGENTS.md |
| 2. Spec | Feature/session scope | Per feature | PRD, architecture docs |
| 3. Source | Per task | Per task | Relevant source files |
| 4. Errors | Per iteration | Per attempt | Test failures, stack traces |
| 5. History | Accumulates | Session | Conversation history |
Principle: Levels 1-2 are curated (high leverage). Levels 3-5 are per-call (keep minimal).
Include ONLY what the agent cannot discover independently:
The agent can discover these itself — including them wastes tokens:
| Context Type | Max Size | Rationale |
|---|---|---|
| AGENTS.md | 500-1000 tokens | ETH Zurich: more = worse |
| SKILL.md (core) | 1000-2500 tokens | Balance detail vs overhead |
| references/ per skill | 500-1000 tokens | Support data, not duplicate |
| System prompt total | < 5K tokens | Beyond this: diminishing returns |
| Anti-Pattern | Problem | Fix |
|---|---|---|
| "Always be thorough" | Forces effort=high, +35% tokens | Remove — model handles this |
| "Think step by step" | Redundant with adaptive thinking | Remove on modern models |
| Repeating the same rule 3x | Token waste, no benefit | State once, clearly |
| Including full API docs | Context overflow | Link to docs, summarize key parts |
| "You are a helpful assistant" | Generic, no value | Use specific task context |
