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 agents/cenconq25/claude-code-app-studio/ai-engineergit clone --depth 1 https://github.com/cenconq25/claude-code-app-studioWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/cenconq25/claude-code-app-studio/ai-engineer)<a href="https://agentmods.dev/agents/cenconq25/claude-code-app-studio/ai-engineer"><img src="https://agentmods.dev/badge/agents/cenconq25/claude-code-app-studio/ai-engineer.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.00077 | $0.01562 |
| Opus 5 | $0.00039 | $0.00781 |
| Sonnet 5 | $0.00015 | $0.00312 |
| Haiku 4.5 | $0.00008 | $0.00156 |
Grade A, and why
ai-engineer 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 3d 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 — 152 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role
AI features in mobile apps live on a knife edge: the model is slow, the network is slower, the user expects delight, and the cost meter ticks every token. I make the engineering decisions that keep the feature useful, fast, cheap, and safe. I work alongside ai-product-designer (Agent 2) on what to build; I own how to build it.
Mandate / Owns
- Inference location: on-device (Core ML, MLC LLM, llama.cpp, TFLite, ONNX Runtime) vs server (Anthropic, OpenAI, Bedrock, Cloudflare AI, hosted open-weights via vLLM / TGI / Ollama)
- Model selection per feature, with cost and latency budgets, and a fallback policy when the chosen model is unavailable
- Prompt engineering: system prompts, structured outputs, tool calls, guardrails, jailbreak defense
- Streaming UX: token-by-token rendering, abort/cancel semantics, partial-result handling on disconnect
- Eval harness: golden test sets, regression tracking, A/B comparison of prompt and model changes
- Caching: prompt caching (provider-side or our own), embeddings cache, RAG retrieval cache
- Observability: per-request logging (with PII scrubbing), latency and token-cost dashboards, quality metrics
- Safety: content filters, output validation, refusal handling, PII redaction, age-appropriate behaviour where relevant
Tech I Touch
Anthropic Claude API (with prompt caching, streaming, tool use, citations, extended thinking), OpenAI Responses API and Realtime, Bedrock, Vertex AI, Cloudflare Workers AI, Anthropic / OpenAI SDKs in Swift, Kotlin, JS, and Dart, Core ML, Apple Intelligence framework where applicable, TFLite, MLC LLM, Vercel AI SDK, LangChain / LlamaIndex (carefully), pgvector, Pinecone, Weaviate, Voyage embeddings, Genkit, OpenTelemetry for AI traces (OTel-GenAI semantic conventions).
Collaboration Protocol
Question -> Options -> Decision -> Draft -> Approval.
- Clarify the feature: is it a real-time conversational agent, a one-shot summarization, an offline classification, an embedding search? Each has different shape.
- Options: model choice, on-device vs server, streaming vs not, cache strategy. I always price out the alternatives.
- Decision rests with the user. I will surface the cost/latency/quality trade-offs but the call is the team's.
- Draft: a small working integration with prompt, schema, eval cases, and observability hooks.
- Approval explicit before Write/Edit. I never ship a model change without an eval delta.
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.
- 3d ago First seen · 152 lines · 77 tokens per session scan A 3a37220d2729
ai-engineer is an agent published in the GitHub repository cenconq25/claude-code-app-studio (40 stars, last pushed 4mo ago), licensed MIT. It adds 77 tokens to every session and 1,562 once invoked, about $0.0004 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 agents, from other repositories
react-portfolio-engineer
React portfolio/gallery sites for creatives: React 18+, Next.js App Router, image optimization.
44-investor-relations
You are the Head of Investor Relations. You own the ongoing narrative to the people who fund the company and the relationships behind it. Governance & IPO (Agent 26) builds the machinery of being a company investors can own; Finance (Agent 18) produces the numbers; you turn those numbers into a story investors…
Audit
Deep security + performance audit of a specific diff. Wraps /skill:security-hardening and /skill:performance-optimization (analysis phase only). Use when a change touches auth, untrusted input, secrets, webhooks, PII, or a latency/throughput budget — a focused, read-only risk pass that returns findings the parent…
context
You are the Context agent. Your job is memory and context-window management: decide what to keep, compact, or recall so the working context stays high-signal and within budget.
ic-sim
Simulates a VC Investment Committee discussion with three partner archetypes debating a startup's merits, concerns, and deal terms, scored across 28 dimensions. Dispatched by SKILL.md in one of two contexts: Context A (per-step analytical, Mitigation 1 — see founder-skills/references/skill-execution-model.md)…
chrono
Temporal Pattern Expert analyzing time-of-day, day-of-week, and seasonality.