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 skills add humanerd-drew/opencode-drewgent --skill cf-worker-modular-architecturegit clone --depth 1 https://github.com/humanerd-drew/opencode-drewgentWrote 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/skills/humanerd-drew/opencode-drewgent/cf-worker-modular-architecture)<a href="https://agentmods.dev/skills/humanerd-drew/opencode-drewgent/cf-worker-modular-architecture"><img src="https://agentmods.dev/badge/skills/humanerd-drew/opencode-drewgent/cf-worker-modular-architecture.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 5.1 | $0.00048 | $0.01862 |
| Opus 5 | $0.00024 | $0.00931 |
| Sonnet 5 | $0.00010 | $0.00372 |
| Haiku 4.5 | $0.00005 | $0.00186 |
Grade A, and why
cf-worker-modular-architecture scanned grade A with 1 finding 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- One change at a time — make the change, watch for wrangler auto-reload, verify with curl, then proceed. How it starts
The opening of the file, as written. The whole thing — 145 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cloudflare Worker Modular Architecture
When a Worker grows beyond 300-500 lines, refactor into: Router → Controllers → Engine → Utilities
Workflow Protocol
- Never commit without user review. Stage changes (
git add -A) and present the diff or summary. Wait for explicit approval beforegit commit. - Run
npm run devand verify the app works after each batch of changes. - One change at a time — make the change, watch for wrangler auto-reload, verify with curl, then proceed.
- When dev server crashes (port conflict / SQLITE_BUSY): kill old process, clear
.wrangler/state/v3/d1/, restart.
Layers
- worker.ts (router, ~100-200 lines): URL matching → dispatch to controller. No business logic.
- controllers/: export
handle*(request, env, url): Promise<Response>. Single responsibility. - engine/ or analysis/: Pure computation, no HTTP, deterministic.
- utils/: Shared helpers (llm, crypto, cors, email)
- data/: Static JSON master data
Key Rules
Controller Pattern
- All controllers share the same signature
(request, env, url): Promise<Response> - Export individual handler functions, not class instances
- Import shared utilities from
../utils/
External API Isolation (PDC Principle)
- Never recalculate what an external API already provides. Pipe it through and transform format only.
- Example: PersonalDateCalculator returns
tenGods,daewoon.direction,wolun[]. Use those values directly rather than recalculating from raw pillars. - The external API is the single source of truth for core domain calculations. Your engine should only compute what the API doesn't provide.
- Pitfall: Duplicating calculation logic creates maintenance burden and subtle drift when the external API gets bug fixes.
LLM Caller Consolidation
- Don't duplicate LLM call logic across controllers. One
utils/llm.tswith:callDeepSeek()— primarycallNvidiaWithFallback()— 3-key fallback (NIM_KEY → FALLBACK → FALLBACK_2)callLLMJson()— DeepSeek → NVIDIA + JSON extraction- 28s per-key timeout to stay under Cloudflare's 30s wall limit
- Each LLM endpoint (report generation, polish, analysis) builds its own system prompt but calls through the same utility.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 145 lines · 48 tokens per session scan A cf483e1745fc
cf-worker-modular-architecture is a skill published in the GitHub repository humanerd-drew/opencode-drewgent (2 stars, last pushed 1mo ago), licensed MIT. It adds 48 tokens to every session and 1,862 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
architecture-paradigm-microservices
Applies microservices for independent deployment and per-service scaling. Use when teams need autonomous release cycles with distinct capability scaling needs.
architecture-paradigm-serverless
Applies serverless FaaS patterns for event-driven workloads. Use when designing bursty workloads with minimal infrastructure and pay-per-execution cost model.
terraform-infrastructure
Structures, writes, and reviews Terraform infrastructure code. Covers module layout, remote state, workspace strategy, variable and secrets handling, CI plan/apply pipeline, naming conventions, and multi-region deployment patterns (provider aliases, per-region state, failover strategies), while delegating shared risk…
health-check-endpoints
Implements liveness and readiness health check endpoints following Kubernetes probe conventions. Covers response schema, dependency checks, Kubernetes probe config, and circuit breaker integration. Invoked when the user asks to add health checks, implement a /health endpoint, or set up Kubernetes probes.
serverless-architecture
Designs and implements serverless functions: function boundary design, runtime selection, event sources, IAM least privilege, configuration, observability, and cost estimation. Invoked when the user asks to build a serverless function, design a Lambda-based system, or move workloads to serverless.
External Dependencies & Vendor Risks
Ensure designs document third-party dependencies, SLAs, quotas, failure modes, and vendor lock-in mitigation.