opensite-ai/opensite-skills / amazon-bedrock
amazon-bedrock
FlaggedBuilds 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 Bedrock models (Claude, Llama, Nova, Titan). Also for prompt caching, quota and throttling diagnosis, cost tracking, migrating between Claude model generations, chunking strategies, and Bedrock model selection. 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. For SageMaker model discovery, customization, or endpoints, use aws-ai-ml.
references/bedrock-agents-to-agentcore-harness/eligibility.md
View on GitHub# Eligibility rubric
Check each condition against the discovery manifest. A **hard-stop** means the agent has a feature with no validated AgentCore Harness path — stop the migration, name the failing condition, and suggest the manual alternative. Do not migrate the eligible parts of an ineligible agent: a half-migrated agent is worse than a clear "not yet."
State the result of *every* condition to the user, not just the first failure — they need the full picture to decide what to fix.
For an **inline agent**, evaluate these same conditions against the fields of the `InvokeInlineAgent` payload (the manifest built from it) rather than a stored agent — e.g. multimodal input from the payload's model/inputs, collaboration/orchestration from its config. The rubric is identical; only the source of the fields differs.
## Hard-stop conditions
### 1. Multimodal input
**Signal:** the agent processes images or audio (vision model for image input, or documented image/audio handling). The validated harness path is text-only.
**Alternative:** keep on Bedrock until a multimodal path is validated. (This is a true hard-stop — don't offer a degraded "migrate anyway with image/audio dropped" path, which would contradict the hard-stop rule above.)
### 2. Multi-agent collaboration
**Signal:** `agentCollaborationMode` is `SUPERVISOR` or `SUPERVISOR_ROUTER`. The supervisor is out of scope — do **not** flatten collaborators into its prompt, wire them as sub-agents, or migrate it alone (dangling references).
**Alternative:** each collaborator is an ordinary Bedrock agent; any that doesn't itself collaborate can be migrated on its own (run this skill once per collaborator). Only the supervisor layer is excluded.
### 3. Unreachable knowledge base
**Signal:** an associated KB is in an account/region the credentials cannot reach — so `bedrock-agent-runtime:Retrieve` can't reproduce its retrieval.
**Alternative:** grant cross-account/region access, then re-run.
**Not the KB *type*:** every type (`VECTOR`/`MANAGED`/`KENDRA`/`SQL`) is reachable via `Retrieve` and eligible; type only picks the wiring (see [mapping.md](references/bedrock-agents-to-agentcore-harness/mapping.md)).
### 4. Custom orchestration
**Signal:** `orchestrationType` is `CUSTOM_ORCHESTRATION` (custom orchestration Lambda). The harness runs its own loop; that control flow has no equivalent and would be silently dropped.
**Alternative:** re-express the logic as harness tools/prompt as a fresh design, or keep on Bedrock.
## Eligible — do not mistake these for blockers
- **DRAFT-only agent** (no published version) — common; treat DRAFT as the source (see [discovery.md](references/bedrock-agents-to-agentcore-harness/discovery.md)).
- **Mixed action-group schema styles** (`functionSchema` and OpenAPI in one agent).
- **Managed KB with non-default retrieval config, code interpreter, session memory** — all have harness targets (see [mapping.md](references/bedrock-agents-to-agentcore-harness/mapping.md)).
- **An agent with a guardrail** — eligible overall; the guardrail capability just can't be carried.
(Eligible to migrate, but whose *capability* the harness can't reproduce — classify **cannot** and surface to the user: **Return-of-Control action groups** and the **guardrail**. See [mapping.md](references/bedrock-agents-to-agentcore-harness/mapping.md).)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |