amazon-bedrock

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.

Quality

94%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Medium

Suggest reviewing before use

The canonical home for this skill is amazon-bedrock in aws/agent-toolkit-for-aws

Quality

Content

88%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A strong, highly actionable routing skill: every section defers detail to real, verified reference files while surfacing Bedrock-specific gotchas (quota reservation mechanics, prepare-agent, inference-profile requirements) that generic AWS knowledge would miss. The main weaknesses are mild redundancy — the maxTokens warning and AgentCore routing each repeated across several sections — and a two-level-deep nested reference bundle for the migration guide.

Suggestions

Consolidate the maxTokens quota warning to the Critical Warnings section (or a single quota reference) and have other sections link to it, instead of restating the full explanation in the Converse-vs-InvokeModel, two Examples, ThrottlingException, and Deploy sections.

Merge the three overlapping routing surfaces (the 'Which Bedrock Capability Do You Need?' table, the AgentCore Services table, and the 'Deploy an agent to AgentCore' prose) into one routing table to remove duplicated guidance.

Flatten or inline the migration guide's nested sub-references (references/bedrock-agents-to-agentcore-harness/{cli,deploy,discovery,eligibility,mapping}.md) so all references are one level deep from SKILL.md, or note in the body that these are sub-pages of the migration guide so navigation stays predictable.

Dimension	Reasoning	Score
Conciseness	Efficient overall — dense tables, service-specific gotchas ("unset values default to the model's maximum and silently reserve far more quota than needed"), and no explanations of concepts Claude already knows. Not score 5 because the maxTokens warning is repeated in 4+ sections and routing is spread across three overlapping tables (capability table, AgentCore services table, Deploy-an-agent prose); not score 3 because nearly every section carries non-obvious, load-bearing information.	4 / 5
Actionability	Fully executable commands throughout: `aws bedrock list-foundation-models --region us-east-1`, a complete converse invocation with model-id and inference-config, `Config(retries={"max_attempts": 5, "mode": "adaptive"})`, plus concrete diagnoses like "Run prepare-agent after ANY configuration change." Not score 4 because commands are copy-paste ready and cover the common cases, including exact error-message remediations.	5 / 5
Workflow Clarity	Sequences are explicit with validation checkpoints: create → data source → ingest → "verify with Retrieve", "polling get-harness until status READY", "check for cacheReadInputTokens", and the 5-point mandatory coverage list for Harness answers. Not score 4 because validation/feedback loops are explicit, and destructive operations carry a confirm-before-proceeding constraint.	5 / 5
Progressive Disclosure	Good structure: a TOC, routing tables with well-signaled links to 24 reference files (all verified to exist on disk), and a clear MCP-vs-local file-resolution guardrail. Not score 5 because the migration path nests two levels deep (SKILL.md → migrate-bedrock-agents-to-agentcore-harness.md → bedrock-agents-to-agentcore-harness/{cli,deploy,discovery,eligibility,mapping}.md) rather than the one-level-deep ideal; not score 3 because all references are clearly signaled and appropriately split.	4 / 5
	Total	18 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

An excellent description: third-person, comprehensive, dense with concrete API names, natural trigger phrases including exact error strings, and explicit exclusions. The only minor criticism is mild redundancy (guardrails and model selection each appear twice across the "Applies when" and "Also for" sentences), which pads length without adding trigger value.

Dimension	Reasoning	Score
Specificity	Lists multiple specific concrete capabilities with named APIs and services: "model invocation (Converse API, InvokeModel), RAG with Knowledge Bases, Bedrock Agents, Guardrails, and AgentCore (including the Harness managed agent loop)" — comprehensive and fully concrete, matching the top anchor; not score 4 because coverage is broad with no significant gaps.	5 / 5
Completeness	Explicitly answers both: what — "Builds generative AI applications on Amazon Bedrock" with a full capability list; when — "Applies when invoking models, setting up Knowledge Bases, creating agents…" with concrete trigger phrases. Not score 4 because the "when" is fully explicit rather than merely present.	5 / 5
Trigger Term Quality	Comprehensive natural-language triggers users would actually say: "invoking models", "setting up Knowledge Bases", "creating agents", "applying guardrails", "deploying to AgentCore", "troubleshooting Bedrock errors (ThrottlingException, AccessDeniedException)", plus synonyms ("migrating/porting/converting") and concrete named targets (Claude, Llama, Nova, Titan, Coinbase CDP, Stripe Privy). Not score 4 because even synonyms and exact error strings users would paste are included.	5 / 5
Distinctiveness Conflict Risk	Clear niche (Amazon Bedrock) with distinct triggers and an explicit exclusion clause "NOT for custom model training, Rekognition, or Comprehend" that prevents overlap with adjacent AWS skills. Not score 4 because the disambiguation is explicit rather than implied, minimizing conflict risk.	5 / 5
	Total	20 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: aws/agent-toolkit-for-aws
Path: plugins/aws-core/skills/amazon-bedrock/SKILL.md
Commit: b8171ad

Reviewed: 5 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.