Judgement, not recall. 400 practice questions across all five domains:
www.claudecertifiedarchitects.com
Judgement, not recall. 400 practice questions across all five domains:
www.claudecertifiedarchitects.com
400 practice questions across all five domains, every answer explained.
www.claudecertifiedarchitects.com
400 practice questions across all five domains, every answer explained.
www.claudecertifiedarchitects.com
Free diagnostic, scored by domain, no signup:
www.claudecertifiedarchitects.com/diagnostic/
Free diagnostic, scored by domain, no signup:
www.claudecertifiedarchitects.com/diagnostic/
400 practice questions across all five domains, every answer explained, free diagnostic to start.
www.claudecertifiedarchitects.com
400 practice questions across all five domains, every answer explained, free diagnostic to start.
www.claudecertifiedarchitects.com
The Claude Certified Architect Foundations exam tests production judgement: when to escalate, how to scope tool access, what breaks in long-context agents. Practice here:
www.claudecertifiedarchitects.com
The Claude Certified Architect Foundations exam tests production judgement: when to escalate, how to scope tool access, what breaks in long-context agents. Practice here:
www.claudecertifiedarchitects.com
www.claudecertifiedarchitects.com
www.claudecertifiedarchitects.com
The Claude Certified Architect Foundations exam tests production judgement: when to escalate, how to scope tool access, what breaks in long-context agents.
www.claudecertifiedarchitects.com
The Claude Certified Architect Foundations exam tests production judgement: when to escalate, how to scope tool access, what breaks in long-context agents.
www.claudecertifiedarchitects.com
The weights matter more than most people expect: Agentic Architecture 27%, Claude Code 20%, Prompt Engineering 20%, Tool Design and MCP 18%, Context Management 15%.
www.claudecertifiedarchitects.com
The weights matter more than most people expect: Agentic Architecture 27%, Claude Code 20%, Prompt Engineering 20%, Tool Design and MCP 18%, Context Management 15%.
www.claudecertifiedarchitects.com
Find out where you stand:
www.claudecertifiedarchitects.com/diagnostic/
Find out where you stand:
www.claudecertifiedarchitects.com/diagnostic/
400 scenario-based practice questions across all five exam domains, every answer explained. 55 free to try, no card.
www.claudecertifiedarchitects.com/cca-practice...
400 scenario-based practice questions across all five exam domains, every answer explained. 55 free to try, no card.
www.claudecertifiedarchitects.com/cca-practice...
One had been telling candidates since April that the correct answer always enforces a constraint outside the prompt!
www.claudecertifiedarchitects.com/blog/guarant...
One had been telling candidates since April that the correct answer always enforces a constraint outside the prompt!
www.claudecertifiedarchitects.com/blog/guarant...
www.claudecertifiedarchitects.com/blog/guarant...
www.claudecertifiedarchitects.com/blog/guarant...
With core.autocrlf=true that writes CRLF over an LF blob. Still modified, still zero/zero.
Use git -c core.autocrlf=false checkout --, and verify with cmp against git show HEAD:, not git status.
With core.autocrlf=true that writes CRLF over an LF blob. Still modified, still zero/zero.
Use git -c core.autocrlf=false checkout --, and verify with cmp against git show HEAD:, not git status.
That's roughly how detection works now. Not "is this true" but "does a person type like this."
That's roughly how detection works now. Not "is this true" but "does a person type like this."
www.claudecertifiedarchitects.com/blog/zero-is...
www.claudecertifiedarchitects.com/blog/zero-is...
The line that stuck: a prompt is a wish addressed to a probabilistic system. Architecture is a guarantee that holds regardless.
habr.com/ru/articles/...
The line that stuck: a prompt is a wish addressed to a probabilistic system. Architecture is a guarantee that holds regardless.
habr.com/ru/articles/...
www.claudecertifiedarchitects.com/blog/how-to-...
www.claudecertifiedarchitects.com/blog/how-to-...
Upstream, the orchestrator gets [] and reports that nothing matched. Confidently. Nobody finds out for a week...
www.claudecertifiedarchitects.com/blog/mcp-too...
Upstream, the orchestrator gets [] and reports that nothing matched. Confidently. Nobody finds out for a week...
www.claudecertifiedarchitects.com/blog/mcp-too...
An agent's account of what it did is not evidence that it did it.
www.claudecertifiedarchitects.com/blog/coding-...
An agent's account of what it did is not evidence that it did it.
www.claudecertifiedarchitects.com/blog/coding-...
Claude read it, inferred something transient, retried.
Your error strings are model input:
www.claudecertifiedarchitects.com/blog/mcp-too...
Claude read it, inferred something transient, retried.
Your error strings are model input:
www.claudecertifiedarchitects.com/blog/mcp-too...
Tool design and MCP is 18% of CCAR-F. Check what it tests:
www.claudecertifiedarchitects.com/diagnostic/
Tool design and MCP is 18% of CCAR-F. Check what it tests:
www.claudecertifiedarchitects.com/diagnostic/
Wrote up what the blueprint actually asks for in each:
www.claudecertifiedarchitects.com/blog/cca-fou...
Wrote up what the blueprint actually asks for in each:
www.claudecertifiedarchitects.com/blog/cca-fou...
Ten questions, scored across the five blueprint domains, no account:
www.claudecertifiedarchitects.com/diagnostic/
Ten questions, scored across the five blueprint domains, no account:
www.claudecertifiedarchitects.com/diagnostic/
The agent read "Conflict", inferred something transient, and retried eleven times in ninety seconds.
The agent read "Conflict", inferred something transient, and retried eleven times in ninety seconds.