--- title: "Amazon Bedrock: review, setup & limitations · Undominated.ai" canonical: https://undominated.ai/skills/aws-amazon-bedrock/ description: "Bedrock model invocation, Knowledge Bases, Guardrails and AgentCore guidance, with quota, prompt-caching and cost-tracking workflows." --- # Amazon Bedrock: review, setup & limitations · Undominated.ai > Bedrock model invocation, Knowledge Bases, Guardrails and AgentCore guidance, with quota, prompt-caching and cost-tracking workflows. [← Explore all skills](/skills/) AI APPLICATION DEVELOPMENT / Amazon Web Services # Amazon Bedrock Bedrock model invocation, Knowledge Bases, Guardrails and AgentCore guidance, with quota, prompt-caching and cost-tracking workflows. [See setup guidance ↓](#setup)[Original source ↗](https://github.com/aws/agent-toolkit-for-aws/tree/c34fbdfde6779593f651f2c85629a85d27231498/skills/core-skills/amazon-bedrock) SOURCE REVIEW Reviewed 2026-10-13 Evidence 5 linked sources Publisher Amazon Web Services Licence [Apache-2.0 ↗](https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/LICENSE) Revision c34fbdfde677 [Read what was—and wasn’t—checked ↓](#review) “Builds generative AI applications on Amazon Bedrock. Covers model invocation (Converse API, InvokeModel), RAG with Knowledge Bases, Bedrock Agents, Guardrails, and AgentCore (including the Harness managed agent loop). Applies when invoking models, setting up Knowledge Bases, creating agents, applying guardrails, deploying to AgentCore, migrating/porting/converting a Bedrock Agent (including inline agents) to an AgentCore Harness, troubleshooting Bedrock errors (ThrottlingException, AccessDeniedException), or choosing models (Claude, Llama, Nova, Titan). Also for prompt caching, quota and throttling diagnosis, cost tracking, migrating between Claude model generations (4.5 to 4.6 to 4.7), chunking strategies, API selection (Converse vs InvokeModel), guardrail capabilities, and model selection. Also covers AgentCore Payments (x402, microtransactions, Payment Manager, Connector, Instrument, Coinbase CDP, Stripe Privy, paid endpoints, agent payments). NOT for custom model training, Rekognition, or Comprehend.” [Amazon Web Services · upstream description ↗](https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/skills/core-skills/amazon-bedrock/SKILL.md) Our analysis follows below. 01 / THE REASONING ## Why this made the selection. - It requires setting maxTokens explicitly on every Converse or InvokeModel call, because an unset value defaults to the model maximum and reserves far more quota, a cause of ThrottlingException named in the file. - It treats the cross-region inference profile prefix as a data residency decision, and asks for time range, grouping and attribution method before writing Cost Explorer queries. ### A good fit for - Diagnosing Bedrock throttling or access errors - Setting up prompt caching or cost tracking for Bedrock model calls ### Weigh up before choosing - Most procedures live in the references directory, which was not opened; the file says to read the relevant reference before responding. - It defaults RAG to the Managed Knowledge Base and new agent builds to AgentCore Harness, using the other options only when the user asks. The install commands do not pin this revision. - MCP retrieval is reserved for the skill package files; the source requires user data and session artifacts to remain in the working directory and prohibits customer data through retrieve_skill. 02 / THE REVIEW RECORD ## What we actually inspected. Source review has boundaries. A clear record is more useful than a “safe” badge. ### Material inspected - skills/core-skills/amazon-bedrock/SKILL.md (frontmatter, file-location guardrail, critical warnings, workflows, troubleshooting, model selection) - plugins/aws-core/skills/amazon-bedrock/SKILL.md (compared byte for byte with the skills/ copy) - README.md (Claude Code, Codex, Kiro and fx installation) - LICENSE Apache-2.0 text ### Our findings - Bedrock has separate API endpoints and the file names a wrong endpoint as a common cause of errors. - Agents need prepare-agent after any configuration change, per the file. ### Not established by this review - The installer was not run, and the skill was not executed against a project or service. No AWS API was called. - The references/ files, assets and scripts were not opened. - Not executed: it was not installed or run. The review applies to the material and revision named here. A newer upstream release can change its behavior. 03 / PUT IT TO WORK ## Add a skill to your workflow. [Upstream setup instructions ↗](https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/README.md) DOCUMENTED COMMAND /plugin install aws-core@claude-plugins-official Copy command ↗ Copying does not execute this command. It may retrieve a newer version than the reviewed source. - The README installs the aws-core plugin from the claude-plugins-official marketplace. In this repository plugins/aws-core/skills/amazon-bedrock/SKILL.md is byte-identical to skills/core-skills/amazon-bedrock/SKILL.md; the copy served by that marketplace was not compared. The README gives npx skills add aws/agent-toolkit-for-aws/skills for Kiro. - Neither path pins revision c34fbdfde6779593f651f2c85629a85d27231498. Compare the installed SKILL.md and references with that revision. - The README says each plugin also bundles the AWS MCP Server configuration. ### Before you start - Claude Code for the plugin path - AWS CLI v2 with credentials and Bedrock model access in the target region ### Compatibility Agent Skills-compatible hosts · Claude Code · Codex · Cursor · Kiro · fx · AWS CLI v2 ### AWS account API calls through the CLI - Runs AWS CLI commands against Bedrock, AgentCore, Service Quotas, CloudWatch and Cost Explorer with the user's credentials. - The file requires an overview before execution and an explanation before each command, and says to confirm before destructive or irreversible operations. ### Cost model Source is Apache-2.0. Bedrock model calls, Knowledge Bases and AgentCore are billed by AWS. Those charges were not measured. 04 / FOLLOW THE EVIDENCE ## The source trail. Our notes are separate from the original resource. Check upstream before adopting a new version. - [Skill definition ↗](https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/skills/core-skills/amazon-bedrock/SKILL.md) Checked 2026-10-13 https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/skills/core-skills/amazon-bedrock/SKILL.md Supports: summary, upstreamDescription, whySelected, limitations, access, review - [Install instructions and skill index ↗](https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/README.md) Checked 2026-10-13 https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/README.md Supports: install, compatibility - [Repository licence ↗](https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/LICENSE) Checked 2026-10-13 https://github.com/aws/agent-toolkit-for-aws/blob/c34fbdfde6779593f651f2c85629a85d27231498/LICENSE Supports: license - [2026-10-15 W26 source re-review at the existing pin ↗](https://raw.githubusercontent.com/aws/agent-toolkit-for-aws/c34fbdfde6779593f651f2c85629a85d27231498/skills/core-skills/amazon-bedrock/SKILL.md) Checked 2026-10-15 https://raw.githubusercontent.com/aws/agent-toolkit-for-aws/c34fbdfde6779593f651f2c85629a85d27231498/skills/core-skills/amazon-bedrock/SKILL.md Supports: upstreamDescription, review.notTested, limitations - [2026-10-15 W26 source re-review at the existing pin ↗](https://raw.githubusercontent.com/aws/agent-toolkit-for-aws/c34fbdfde6779593f651f2c85629a85d27231498/README.md) Checked 2026-10-15 https://raw.githubusercontent.com/aws/agent-toolkit-for-aws/c34fbdfde6779593f651f2c85629a85d27231498/README.md Supports: install.command KEEP COMPARING ## Other approaches to consider. Related by category or shared topics. These are alternatives to inspect, not a measured quality order. [### Agent Platform GenAI Inference ↗ Calls Gemini and third-party OpenMaaS models on Google Cloud Agent Platform, asking for model, project and region instead of defaulting.](/skills/google-agent-platform-inference/)[### Anthropic Claude API ↗ Routes Claude API work to language-specific SDK guidance and distinguishes self-hosted tool loops from managed agent execution.](/skills/anthropics-claude-api/)[### Datadog LLM Observability Instrumentation ↗ Instruments a Python or Node.js/Next.js backend with Datadog LLM Observability, auditing session-ID plumbing and persisting the dependency before reporting.](/skills/datadog-labs-dd-instrument-llmo/) [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/) ## Continue your investigation - [Choose a toolkit](/tools/) - [Inspect execution hosts](/agents/) - [Check tool access](/mcp-servers/) - [Check project dependencies](/check-repo/)