Zaguán Blade + zcoderd
Evidence-grounded AI engineering inside your codebase.
Zaguán helps software teams use AI for controlled investigation: security review, blindspot discovery, codebase research, visual debugging, and validated change loops with memory and traceability.
Controlled investigation loop
Blade Protocoluser: audit this auth change
zblade: workspace + diff + intent
zcoderd: request files, tests, logs
tools: rg, typecheck, web, visual checks
result: findings + evidence + fix planNot an AI IDE. Not a token reseller.
The business is controlled AI engineering work: investigate the codebase, gather evidence, take action, validate the result, and keep a traceable record of what happened.
Question Engine
Zaguán is built around asking what must be inspected before an answer can be trusted. The useful output is often the missing probe, not the first confident answer.
Evidence Loop
Workspace context, file reads, tool output, visual checks, tests, and web research are treated as evidence. Model judgment is kept separate from what was actually observed.
Controlled Runtime
Blade is the reference workspace. zcoderd is the server-side investigation runtime. The durable product is the protocol, memory, workflow discipline, and audit trail between them.
Zaguán Blade
The workspace where the investigation becomes real.
Zaguán Blade is not positioned as another VS Code fork to win on editor chrome. It is the reference surface for evidence-grounded AI engineering: focused context, controlled tools, persistent memory, and workflows that can explain themselves later.
Blade + zcoderd
A client and runtime pair for serious AI-assisted engineering.
Packaged workflows, not generic prompting.
The strongest Zaguán workflows encode engineering judgment into repeatable investigation methods, then produce outputs that a developer, reviewer, or technical lead can inspect.
Security Review
Investigate trust boundaries, exploit value, dangerous sinks, and false positives with an evidence ledger instead of a loose chat transcript.
Blindspot Discovery
Turn hidden assumptions into concrete probes, then preserve what was found so the team does not rediscover the same risk later.
Visual Debugging
Use controlled visual evidence gathering for UI failures, regressions, and state problems that cannot be trusted from code alone.
Validated Change Loops
Move from investigation to code change to tests and review with the reasoning, commands, and remaining uncertainty kept together.
The model gateway still matters. It is infrastructure.
CoreX supports the investigation runtime with model access, provider translation, routing, and billing. It is valuable because it keeps Zaguán Blade and zcoderd free to choose the right model for the job without turning your product into a provider integration project.
OpenAI-Compatible Access
Keep a familiar SDK path for application calls and automation while Zaguán handles provider-specific translation.
Runtime Control
Route requests, centralize usage, and keep model decisions connected to the engineering workflow that needed them.
Provider Flexibility
The point is not cheap tokens. The point is being able to pick the model that fits the investigation, validation step, or production call.
Subscriptions for Blade-powered engineering work.
Plans include monthly credits for Zaguán Blade, zcoderd workflows, and model calls through the gateway. Pick the amount of controlled investigation and build work you expect to run each month.
Side Project
For individual builders using Blade seriously
- 400 credits/month
- Zaguán Blade workflows
- Security and blindspot probes
- Gateway model access
- Email support
Startup
For small teams running deeper investigations
- 1,000 credits/month
- Zaguán Blade workflows
- Validated change loops
- Gateway model access
- Priority support
- Usage analytics
Scale
For agencies and teams with steady AI engineering work
- 2,500 credits/month
- Zaguán Blade workflows
- Security review and visual debugging
- Gateway model access
- Priority support
- Usage analytics
- Team management
Extra credit packs are available in the dashboard for active subscribers and do not expire.
Use AI where the evidence matters.
Start with Zaguán Blade for controlled engineering investigation. Use the gateway when your application or workflow needs reliable model access underneath it.
Frequently Asked
Is Zaguán still an AI model gateway?
The gateway is still part of the system, but it is not the main business story. It supports the larger product: controlled, evidence-grounded AI engineering workflows.
Is Zaguán Blade the whole product?
Blade is the reference environment where the workflow is easiest to see and use. The deeper product is the loop between Blade, zcoderd, the protocol, tools, memory, validation, and traceable outputs.
Who is this for first?
The first strong fit is developers and small teams doing work where correctness, privacy, and reviewability matter more than quick generation: security review, risky changes, unfamiliar codebases, and hard-to-reproduce UI failures.
Product updates only.
Zaguán Blade changes, workflow releases, protocol updates, and model infrastructure notes. No generic AI commentary.