Open-source infrastructure / Developer infrastructure
Open Source Infrastructure
OSS work around Claude Code workflow discipline, MCP visibility, public API access, and Infisical Agent Vault diagnostics.
A set of developer-infrastructure projects focused on repeatable LLM-based development workflows, MCP observability, no-key public API access, and agent-secret diagnostics.
Status
Live
Timeline
Ongoing OSS and developer-infrastructure work
Domain
Developer infrastructure
Why
Open-source infrastructure
Developer workflow
Spec → Build → PR → Review
Mode
OSS
Focus
AI dev
Goal
Repeatable
Claude Code workflow structure, MCP visibility, public API tools, and Agent Vault diagnostics.
Stack
Languages, services, data sources, and operating pieces behind the build.
Code Proof
What The Build Actually Contains
Projects
4 live
Workflow
1.9k LOC
Wiretap
3k LOC
Public APIs
2.8k LOC
Agent Vault
PR #134
Workflow preview
Spec To PR
01
CLAUDE.md
Repo rules and build conventions
02
AGENTS.md
Agent autonomy boundaries
03
SPEC_TEMPLATE.md
Scope before code
04
PR_TEMPLATE.md
Explain the change
05
WORKFLOW.md
Repeatable loop
06
examples/
Real project starting points
Live OSS work
Four Live Projects
Lead OSS project
Claude Code Workflow
A complete Claude Code workflow: repo rules, agent boundaries, spec-first templates, PR discipline, setup script, and examples for real app types.
View GitHub
MCP observability
MCP Wiretap
A zero-config proxy that wraps any MCP server, forwards stdio unchanged, logs JSON-RPC traffic to SQLite, and supports export, replay, sessions, and search.
View GitHub
No-key agent tools
MCP Public APIs
A zero-config MCP server exposing useful no-key public APIs to agents: weather, crypto, geocoding, HN, NASA, dictionary, Wikipedia, npm, GitHub, DNS, and more.
View GitHub
OSS contribution
Infisical Agent Vault PR
A contribution branch for Agent Vault request-log inspection and diagnostics: CLI log commands, inspect command, diagnostics package, tests, and docs.
View PR
System shape
Developer workflow
Spec → Build → PR → Review
Mode
OSS
Focus
AI dev
Goal
Repeatable
Claude Code workflow structure, MCP visibility, public API tools, and Agent Vault diagnostics.
Implementation
Code Behind The Surface
Claude workflow guardrail
mdThe core move behind the product surface.
Idea or ticket
-> SPEC_TEMPLATE.md
-> Claude Code builds with CLAUDE.md loaded
-> PR_TEMPLATE.md
-> Review and real test
-> MergeProduct Model
tsThe useful answer has to be modeled before the UI can make it obvious.
type ProductSurface = {
input: "Developer infrastructure";
signal: "How do you help developers, especially fast-moving AI builders, move from vibes to a";
proof: string[];
};
const surface: ProductSurface = {
input: "Developer infrastructure",
signal: "Claude Code workflow structure, MCP visibility, public API tools, and Agent Vault di",
proof: [
"Claude Code workflow system",
"MCP Wiretap inspection",
"MCP Public APIs server",
"Infisical Agent Vault diagnostics PR"
],
};Hard Part
tsEvery build has a constraint: data quality, workflow shape, trust, speed, or operational risk.
const constraint = {
project: "Open Source Infrastructure",
status: "Live",
category: "Open-source infrastructure",
hardPart: "The point is better AI-assisted building: inspectable, repeatable, and less chaotic....",
};
shipSurface(constraint);Project Logic
Why This Exists
The point is not to show another screen. It is to show the gap, the build constraint, and the proof of work.
Mission
How do you help developers, especially fast-moving AI builders, move from vibes to a workflow they can repeat and trust?
LLM-based development gets messy fast when there is no spec, no repo-level guidance, no review habit, and no line between what an agent can decide and what a human should own. The same pattern shows up in MCP and agent infrastructure: hidden behavior kills trust.
Build
What Had To Work
I built Claude Code Workflow as the anchor project, then grouped the rest of the OSS work around the same idea: MCP Wiretap for protocol inspection, MCP Public APIs for zero-key agent tools, and an Infisical Agent Vault PR for request-log inspection and diagnostics.
Why It Matters
Spec → Build → PR → Review
Makes AI and MCP development more inspectable, repeatable, and easier to trust.
Hard Parts
AI-assisted builds need rails
A lot of people are building with LLMs before they understand software workflow. Claude Code Workflow gives them a path: plan first, constrain the agent, review the work, test it for real.
Agent tools need inspection
MCP and agent-secret systems are powerful because they act on behalf of the user. That also means logs, diagnostics, and protocol visibility are not nice-to-have features; they are how trust is earned.
OSS has to be immediately usable
The projects are designed around low-friction starts: one command for Wiretap, one command for Public APIs, clear templates for Claude Code Workflow, and concrete CLI diagnostics for Agent Vault.
Decisions
Next Move
I would keep expanding Claude Code Workflow with more example stacks, team operating patterns, and real-world before/after workflows for solo builders and teams.
Tell Me About Your Project
Bring Me The Bottleneck.
I’ll Build The Answer.
Tell me what people are trying to do, where the current path breaks, and what kind of useful answer should exist.
Market Gap
Demand exists, but the answer is missing.
Workflow Drag
The work is still too manual, slow, or scattered.
Product Wedge
A small surface could prove the larger opportunity.