CCAR-F facts to memorise: 30 recall cards
Thirty facts that are pure recall — the ones you either know cold or lose a mark on. Cover the right-hand column and work down the list.
| Prompt | What you need to know |
|---|---|
stop_reason: "tool_use" | Claude wants a tool. Execute it, append a matching tool_result, and continue the loop. |
stop_reason: "max_tokens" | Truncated, not finished. Never parse it as a final answer — JSON in it is likely malformed. |
| Only channel from parent to subagent | The Agent tool's prompt string. Paths, errors and decisions must be written into it explicitly. |
| Four benefits of subagents | Context isolation · parallelisation · specialised instructions · tool restriction. |
| Settings precedence | Managed → command line → project local → shared project → user. |
CLAUDE.md loading | Concatenated root-down, not overridden. Subdirectory files load on demand. |
| Import depth limit | Four hops. Imported files still load at launch, so imports organise but do not save context. |
| Permission rule evaluation | deny beats ask beats allow. Deny cannot be overridden. |
Bash(ls *) vs Bash(ls*) | With the space: ls, ls -la, not lsof. Without: lsof matches too. |
| Hook exit code 2 | Blocks the action; stderr becomes the reason Claude sees. Cannot be overridden by JSON. |
| Hook exit code 0 | Success. JSON on stdout is parsed for a permission decision. |
| MCP tool naming | mcp__<server>__<tool>. mcp__* matches every MCP tool anywhere. |
| MCP server primitives | Tools, resources, prompts. (Client side: sampling, elicitation, logging.) |
| MCP transports | stdio for a local process serving one client; Streamable HTTP for remote, many clients. |
| Project-scoped MCP file | .mcp.json at the repo root — the one you commit. Use ${VAR} for secrets. |
strict: true | Tool names and inputs conform to their schemas. Different from output_config.format. |
output_config.format | JSON Schema constraining Claude's final response. Max 20 strict tools per request. |
| Unsupported in schemas | pattern, allOf, not, if/then/else, dependentSchemas, recursion. |
| Context editing defaults | Trigger at 100,000 input tokens; keep the last 3 tool uses. |
clear_at_least | Minimum cleared per activation, so the cache invalidation is worth its cost. |
| Cache breakpoints | Four maximum, ordered tools → system → messages. Mark the last identical block. |
| Cache pricing | Read 0.1× base. Write 1.25× (5 min TTL) or 2× (1 hour TTL). |
| Batches API | 50% off · ≤100,000 requests / 256 MB · most within 1 h, all within 24 h · no streaming. |
| Retryable HTTP codes | 429, 500, 504, 529. Not 400, 401, 403, 404, 413. |
--bare | Skips auto-discovery of hooks, skills, commands, agents, plugins, MCP, memory, CLAUDE.md. Use in CI. |
Does not survive /compact | The skill listing, and instructions given only in conversation. |
| Subagent default limits | Depth 3, concurrency 20, no spend cap unless you set maxBudgetUsd. |
disable-model-invocation: true | Only a human can invoke the skill. Use it for deploys and destructive commands. |
| Escalation trigger | Rule-based on observable state — never a model-reported confidence score. |
| Passing findings between agents | Structured records with source IDs. Prose flattening loses citations for good. |
← Previous: Questions & answers — Domain 5 | Next: Six-week study plan →
Explore Other Certification Study Guides
Switch to another exam guide to prepare for your next Claude certification:
Claude Certified Associate — Foundations
Entry-level certification for non-coders: prompting, output evaluation, model selection, workflow integration, configuration, governance & responsible use.
Claude Certified Architect — Professional
Advanced enterprise architecture: solution design, system integration, evaluation frameworks, governance, safety & developer enablement.
Claude Certified Developer — Foundations
Build software on Claude: API integration, model selection, tools & MCP servers, agents vs workflows, security, safety & debugging.