Premium skill
Write agent prompts you can defend
Rewrites the prompts your agent actually runs on: a repo-wide audit of contradictions and token waste, prompts as versioned code, and before and after evals proving every change.
Install
Install with skills CLIPremium skills install through a short-lived private link.
$ npx skills add … --skill write-agent-promptsOpen Premium install
Sign in with Premium to generate a fresh install link.
Filesystem
Browse the authored file tree — tools, references, and evals included. Source unlocks with Premium.
SKILL.mdmarkdownUnlock skill source
Browse every authored SKILL.md with premium.
File tree for write-agent-prompts stays visible so you can see the pack before unlocking.
Use this when
- 01An agent that ignored half its instructions after years of prompt accretion.
- 02A 4,000-token monolith pulled into reviewable prompt modules with no behavior change.
- 03Wrong-tool calls fixed by clearer descriptions, not a silent API redesign.
- 04Few-shot examples chosen from real traces and cut when they did not earn their tokens.
What you get
- Inventory the effective prompt
- System text, tool descriptions, schemas, and examples as one document.
- Prompts as code
- Extract strings into owned modules. Prove equivalence before craft rewrites.
- Measure before and after
- Golden baseline first. Keep only slices that improve the gate.
- Frame untrusted content
- Retrieval, tool results, and uploads as data, never as instructions.
Unlock the files
Available with Premium. The private install link expires; the downloaded skill does not. Generate a fresh link when you want to update it.
Other Premium skills
Lifecycle skills and UI craft skills
- PlanPlan AI SDK AgentBefore you write agent code
- BuildAISDK AgentsImplementing agents, chat apps, and streaming UIs
- EvaluateBuild AI SDK Agent EvalsProtecting critical agent journeys before you optimize
- ObserveObserve AI SDK AgentDebugging and attributing agent behavior in production
- OptimizeOptimize AI SDK AgentImproving quality, speed, cost, or reliability with proof
- InstructWrite Prompt ContractsAgents that pick the wrong tool or narrate a plan instead of acting
- Chat UIBetter Chat UIBuilding or reviewing the conversation surface
- Generative UIBetter Generative UITurning model output into interactive product UI
- PrototypeTest UI VariantsSurfaces where one polished mock would hide the real choice
- IngestIngest Messy DocumentsPDF, scan, and table-heavy corpora feeding RAG or extraction
- GroundGround AI SDK AgentMaking an agent answer from your data with proof
- MemoryAgent MemoryAgents that should remember users across sessions
- ExposeBuild MCP ServerMaking your product agent-accessible
- ExtractExtract Structured DataInvoices, resumes, contracts, forms, and free text to typed data
- ShipShip AI SDK on CloudflareGetting an AI SDK agent into production on Cloudflare
- ShipShip AI SDK on AWSGetting an AI SDK agent into production on AWS
- MobileAI SDK on MobileShipping AI SDK chat in an Expo or React Native app
- MigrateMigrate to AI SDKLeaving LangChain, the Assistants API, or hand-rolled LLM code