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 lukedj78/dev-flow --skill project-infra-setupgit clone --depth 1 https://github.com/lukedj78/dev-flowWrote 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/lukedj78/dev-flow/project-infra-setup)<a href="https://agentmods.dev/skills/lukedj78/dev-flow/project-infra-setup"><img src="https://agentmods.dev/badge/skills/lukedj78/dev-flow/project-infra-setup/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/lukedj78/dev-flow/project-infra-setup"><img src="https://agentmods.dev/badge/skills/lukedj78/dev-flow/project-infra-setup.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00219 | $0.01601 |
| Opus 5.5 | $0.00088 | $0.00640 |
| Sonnet 5.5 | $0.00044 | $0.00320 |
| Haiku 4.5 | $0.00022 | $0.00160 |
Grade A, and why
project-infra-setup 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 — 41 lines — stays where its author put it; the contents beside it link to each section on GitHub.
project-infra-setup — the third-party environments, from the CLI
Everything outside the codebase that a project needs before module-add can wire it: Git, Vercel, the connectors (Neon, Resend, Blob), DNS, secrets. Done once per project, in a fixed order, with the traps already written down. Worked on BidFlow (2026-10-03). The commands and findings are in references/provisioning.md; this file is the order and the rules.
It owns the accounts and the CLI-level resources. It does not write application code (module-add owns that), does not deploy (vercel-deploy owns that), and does not touch the PRD, design or tasks.
On Vercel the line with module-add deploy is: this skill creates and links the project, sets the project-level function region, and puts the variables' values in; module-add deploy writes vercel.json (its regions must equal the region set here, read from docs/infrastructure.md), the build settings, and the matrix of which variables each environment needs. It does not link.
Rules (they held up)
- Every outward step is confirmed, one at a time: create the repo, create the project, create a database, write a DNS record, create or revoke a key, send a test email. A "yes" to the plan is not a "yes" to each step.
- Check the region of every service before and after creating it. The default is US (
iad1) for functions, Blob and Neon, and a database region cannot be changed afterwards. Read it back from the service (PGHOST, the project API), not from the CLI's summary. - Secrets never go through chat. The user types them in their own terminal (
vercel env add NAME --cwd <project>), or you generate and set them yourself from stdin without printing. Narrow every key you can (a send-only key limited to the domain), prove it works against the provider's sink address, then revoke the broad one. - Whatever a tool installs on its own is reviewed before it stays. Marketplace integrations drop third-party agent skills into the repo (
.agents/,.claude/skills/,skills-lock.json), every time. Move them out, read them, let the user decide (registry-intake skill-review). They steer an agent toward the vendor's stack. - Read the plan's terms for the use the project really has. Vercel Hobby is restricted to non-commercial personal use; a company tool is arguably commercial. It is the user's decision: record it as a known risk to revisit, do not decide for them and do not keep asking.
- Reuse what exists. Look at
vercel whoami,vercel teams ls,vercel domains ls, the DNS zone and the integrations first. A Resend account already in the zone beat the marketplace route, which collided with it. - Do not connect Git to Vercel before the app exists (the first push would publish the repo's documents at a public URL), and do not assign the custom domain before the first deploy.
- Say what failed and what you got wrong (a database that landed in the wrong region was deleted and recreated), and what you did not look at.
What ships with it
4 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.
- 2d ago Changed 2a49a5fdfb87
- 3d ago Changed · +2 lines 1d3c1f0395a0
- 5d ago Changed f87104d1ee56
- 16d ago First seen · 39 lines · 219 tokens per session scan A ade633630b30
project-infra-setup is a skill published in the GitHub repository lukedj78/dev-flow (6 stars, last pushed yesterday), licensed MIT. It adds 219 tokens to every session and 1,601 once invoked, about $0.0009 per session on Opus 5.5. 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 2026-10-04.
Other skills, from other repositories
moai-platform-database-cloud
Cloud database platform specialist covering Neon (serverless PostgreSQL), Supabase (PostgreSQL 16 with real-time), and Firebase Firestore (NoSQL with offline sync). Use when choosing or setting up cloud databases.
backup-recovery
Design and verify backups and restoration for owned databases, files and application state. Use when preparing recovery, migrating data or testing backups; protect the source and prove restore into an isolated destination before calling recovery ready.
neon-postgres
Expert patterns for Neon serverless Postgres, branching, connection.
cloud-sql-basics
This file generates or explains Cloud SQL resources. Use this file when the user asks to create a Cloud SQL instance or database for MySQL, PostgreSQL, or SQL Server. Cloud SQL manages third-party MySQL, PostgreSQL, and SQL Server instances as resources in Cloud SQL. For example, when Cloud SQL creates an open-source…
aws-rds
Manage Amazon RDS for fully managed relational databases. Provision MySQL, PostgreSQL, and Aurora instances with automated backups and patching. Configure read replicas for scaling, snapshots for recovery, and Aurora Serverless for variable workloads.
blueprint-from-collection-import
Move the copied collection's records into Postgres and the files they point at into Storage, proven on a freshly reset local database.