aws-bedrock-agentcore-skill

Name: aws-bedrock-agentcore-skill
Author: ferdinandobons

2installs29forks

How to install aws-bedrock-agentcore-skill?

ferdinandobons/awsbedrockagentcoreskill/aws-bedrock-agentcore-skill

$npx -y skills add ferdinandobons/AWSBedrockAgentCoreSkill --skill aws-bedrock-agentcore-skill

Installs into the current project.

Summary

Authoritative, source-cited playbook for designing, configuring, deploying, and troubleshooting production-grade AI agents on AWS using the Strands Agents SDK, Amazon Bedrock (Converse API, Guardrails, Knowledge Bases), and Amazon Bedrock AgentCore (Runtime, Memory, Gateway, Iden

Files · 1

View on GitHub

SKILL.md

1	# AWS Bedrock AgentCore Skill
2
3	The definitive, source-cited guide for building AI agents on AWS. This skill does not
4	hand you a single template - it gives a coding agent the official directives, best
5	practices, working snippets, and source URLs to autonomously configure the right agent
6	for the user's specific use case.
7
8	Every claim in this skill is backed by an official source. The source index lives in
9	[references/sources.md](references/sources.md) - open it whenever you need to re-read a
10	primary source or verify a detail before recommending it.
11
12	## How to use this skill
13
14	1. Read the decision tree below and identify the user's use case and required pattern.
15	2. Open only the reference files that match (progressive disclosure - the references are
16	large and detailed; don't load all of them). Each row of the [reference index](#reference-index)
17	says when to open which file.
18	3. Confirm maturity before recommending. Features are labeled GA / Preview. Never propose a
19	Preview feature as a production default - surface it with an explicit warning (see
20	[GA vs Preview](#ga-vs-preview)).
21	4. Re-verify time-sensitive facts. Model IDs, prices, and quotas change. This skill points to
22	the live model cards, the Bedrock pricing page, and the Service Quotas console for exact numbers
23	instead of hard-coding values that rot.
24	5. Cite your sources back to the user. When you make a recommendation, name the official URL
25	it came from so the user (and you) can re-open it.
26
27	## Core principles (apply to every AWS agent)
28
29	These are the cross-cutting rules that hold regardless of pattern. The detailed versions, with
30	sources, are in the reference files.
31
32	- Default to `BedrockModel` / the Bedrock Converse API. Never use the legacy `InvokeModel` API.
33	Converse is the unified, model-agnostic surface; every capability (tool use, prompt caching,
34	guardrails, reasoning/thinking, service tiers) maps onto it. → [references/bedrock.md](references/bedrock.md)
35	- Always set an explicit region. In boto3 / Strands `BedrockModel`, pass `region_name` explicitly,
36	or set `AWS_DEFAULT_REGION`. `AWS_REGION` is the lowest-priority fallback in the boto3 resolution
37	chain (after `region_name`, `AWS_DEFAULT_REGION`, and profile region) - prefer `AWS_DEFAULT_REGION`
38	or pass `region_name` directly to avoid silent misconfiguration.
39	→ [references/strands.md](references/strands.md)
40	- IAM least-privilege with confused-deputy protection. Scope `bedrock:InvokeModel*` to the exact
41	model ARN (never `*` in production), and put `aws:SourceAccount` + `aws:SourceArn` conditions on
42	every service trust policy. → [references/security-iam-cost.md](references/security-iam-cost.md)
43	- Mind the token quota mechanics (Claude 3.7+ / 4.x): at request start, `input_tokens +
44	max_tokens` is reserved 1:1 from the TPM quota; at request end, actual output tokens are billed
45	at 5×. An oversized `max_tokens` over-reserves quota up front, blocking concurrent requests - that
46	is why you should size `max_tokens` to the real need.
47	→ [references/security-iam-cost.md](references/security-iam-cost.md)
48	- Verify model access before deploy, not at runtime. A model you haven't enabled fails the first
49	Converse

People who liked this also viewed

6 picks

Type

aws-bedrock-agentcore-skill

2installs29forks

View on GitHub

How to install aws-bedrock-agentcore-skill?

ferdinandobons/awsbedrockagentcoreskill/aws-bedrock-agentcore-skill

$npx -y skills add ferdinandobons/AWSBedrockAgentCoreSkill --skill aws-bedrock-agentcore-skill

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# AWS Bedrock AgentCore Skill
2
3	The definitive, source-cited guide for building AI agents on AWS. This skill does not
4	hand you a single template - it gives a coding agent the official directives, best
5	practices, working snippets, and source URLs to autonomously configure the right agent
6	for the user's specific use case.
7
8	Every claim in this skill is backed by an official source. The source index lives in
9	[references/sources.md](references/sources.md) - open it whenever you need to re-read a
10	primary source or verify a detail before recommending it.
11
12	## How to use this skill
13
14	1. Read the decision tree below and identify the user's use case and required pattern.
15	2. Open only the reference files that match (progressive disclosure - the references are
16	large and detailed; don't load all of them). Each row of the [reference index](#reference-index)
17	says when to open which file.
18	3. Confirm maturity before recommending. Features are labeled GA / Preview. Never propose a
19	Preview feature as a production default - surface it with an explicit warning (see
20	[GA vs Preview](#ga-vs-preview)).
21	4. Re-verify time-sensitive facts. Model IDs, prices, and quotas change. This skill points to
22	the live model cards, the Bedrock pricing page, and the Service Quotas console for exact numbers
23	instead of hard-coding values that rot.
24	5. Cite your sources back to the user. When you make a recommendation, name the official URL
25	it came from so the user (and you) can re-open it.
26
27	## Core principles (apply to every AWS agent)
28
29	These are the cross-cutting rules that hold regardless of pattern. The detailed versions, with
30	sources, are in the reference files.
31
32	- Default to `BedrockModel` / the Bedrock Converse API. Never use the legacy `InvokeModel` API.
33	Converse is the unified, model-agnostic surface; every capability (tool use, prompt caching,
34	guardrails, reasoning/thinking, service tiers) maps onto it. → [references/bedrock.md](references/bedrock.md)
35	- Always set an explicit region. In boto3 / Strands `BedrockModel`, pass `region_name` explicitly,
36	or set `AWS_DEFAULT_REGION`. `AWS_REGION` is the lowest-priority fallback in the boto3 resolution
37	chain (after `region_name`, `AWS_DEFAULT_REGION`, and profile region) - prefer `AWS_DEFAULT_REGION`
38	or pass `region_name` directly to avoid silent misconfiguration.
39	→ [references/strands.md](references/strands.md)
40	- IAM least-privilege with confused-deputy protection. Scope `bedrock:InvokeModel*` to the exact
41	model ARN (never `*` in production), and put `aws:SourceAccount` + `aws:SourceArn` conditions on
42	every service trust policy. → [references/security-iam-cost.md](references/security-iam-cost.md)
43	- Mind the token quota mechanics (Claude 3.7+ / 4.x): at request start, `input_tokens +
44	max_tokens` is reserved 1:1 from the TPM quota; at request end, actual output tokens are billed
45	at 5×. An oversized `max_tokens` over-reserves quota up front, blocking concurrent requests - that
46	is why you should size `max_tokens` to the real need.
47	→ [references/security-iam-cost.md](references/security-iam-cost.md)
48	- Verify model access before deploy, not at runtime. A model you haven't enabled fails the first
49	Converse

People who liked this also viewed

6 picks

Type