Web Search Tool on Amazon Bedrock AgentCore

# Fix list: Web Search Tool on Amazon Bedrock AgentCore

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/agentcore-web-search, the October 2026 research run, assessed 17 October 2026. Grade BB, 74.3 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Web Search Tool on Amazon Bedrock AgentCore: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Payments & pricing, 30 out of 100, up to 8.8 more on the total

Why it scored 30: No x402, MPP or L402 on the Gateway or the tool. AgentCore payments is a separate component for an agent paying third parties (0). Per-query price published without a login, $7.00 per 1,000, with a worked example that adds Gateway invocations at $0.005 per 1,000 (20). No Web Search allowance, but new AWS accounts get up to $200 in Free Tier credits and the Free Tier FAQ says most new customers need no payment method, so half, as on the other AWS listings (10). A person signs up in a browser and sets up a gateway, a target and IAM (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 2. Agent ergonomics, 69 out of 100, up to 5 more on the total

Why it scored 69: One tool with a compact schema, and results are snippets AWS says are extracted for model context. Each result comes as a JSON string in a text block and again as `structuredContent` (22 of 25). `maxResults` 1 to 25, per-request domain include and exclude lists and published-date bounds from 1.2.0. No pagination, offset, country or language controls (15 of 20). JWT failures return 401 or 403 with a `WWW-Authenticate` header naming the scopes, and throttling returns `retryAfter`, but no error codes for the tool itself are documented (10 of 20). Searches only read, so retries are safe apart from the per-query charge. We did not see `tools/list` output, so annotations such as `readOnlyHint` are unconfirmed (12 of 20). Only `query` is required, but the default connector 1.1.0 lacks the filters, and a first call needs a gateway, a target, a service role and inbound auth. Any MCP client calls it, and set-up runs through AWS SDKs in many languages (10 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 3. Schema & documentation, 78 out of 100, up to 3.6 more on the total

Why it scored 78: The tool's `inputSchema` is published in full as JSON Schema, with `query` required, and the Gateway control plane is in AWS's service models (25). The developer guide has an llms.txt of 993 lines and serves every page as Markdown (10). The page explains purpose, use cases, the two filter layers and how they compose, but says nothing on when another tool fits better (12 of 20). Types are given, but the 200-character limit on `query`, the 1 to 25 range of `maxResults` and the ISO-8601 date format sit in descriptions, with no `maxLength`, `minimum`, `maximum` or `format` (9 of 15). Python, AWS CLI and AgentCore CLI set-up examples and one response example. No error responses for the tool are documented, and the June release note links to a page that now redirects to the guide's front page (7 of 15). Semver connector versions with release dates, a lifecycle table, monthly release notes and dated What's New posts (15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 4. Security & auth, 85 out of 100, up to 2.6 more on the total

Why it scored 85: Gateway inbound auth is OAuth JWT from the owner's identity provider with allowed scopes advertised in the 401 challenge, or IAM SigV4 scoped to one gateway ARN, and the agent never holds a search credential (30). The tool only reads. The owner's target-level include and exclude lists cannot be relaxed by the agent, the service role's `InvokeWebSearch` names one service-owned ARN, and Cedar policies on the gateway can forbid the tool per caller. A no-authorisation gateway mode exists, with AWS's warning against production use (17 of 20). Results are third-party page text and the Web Search pages give no guidance on treating them as untrusted. Gateway guardrail policies can screen inputs for prompt attacks and suppress outputs, documented in general and not for this tool (7 of 15). CloudWatch metrics by method and tool name, and CloudTrail data events for `InvokeGateway` when the owner turns them on, at extra cost (13 of 15). A vulnerability disclosure programme on HackerOne, a security.txt renewed to expire on 16 October 2027, and SOC and ISO 27001 among the programmes the AgentCore FAQ lists. No paid bounty found, so 2 held back as on every AWS listing (18 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 5. Reliability, 90 out of 100, up to 2 more on the total

Why it scored 90: Graded on the hosted lines, for the Gateway endpoint and the connector agents call. AWS Health Dashboard with per-service, per-Region history (20). The dashboard's history file, read on 17 October 2026, names AgentCore in one event in the 90 days, 2 hours 40 minutes of packet loss in one zone of eu-south-2 on 4 October 2026, marked informational and in a Region where Web Search is not sold. Two us-west-2 events, on 24 July and 3 September, don't name AgentCore. Read as clean, as on the Memory listing (30). Quotas published, 10 Web Search requests a second and 200 tool calls a second per gateway and per account (15). Owner-set gateway rate limits return an error body with `retryAfter` in seconds and searches have no side effects, but the response to the service's own Web Search quota is not documented (10 of 15). The AgentCore FAQ says the Amazon Bedrock SLA applies, but that SLA, last updated 4 October 2023, does not name AgentCore (5 of 10, as on the Code Interpreter listing). Generally available since 16 June 2026 (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 6. Transparency & trust, 77 out of 100, up to 2 more on the total

Made of editorial 60, provenance 93.

Why it scored 77: Closed service under the AWS Customer Agreement and Service Terms, updated 13 October 2026, with an acceptable-use section on the docs page (15 of 30). The docs say queries are served inside AWS and not sent to a third-party engine, the Gateway data protection notes say data is encrypted with a service key or a customer managed key, and section 50.3 of the Service Terms does not list AgentCore for content use. No retention period for queries was found (15 of 30). The connector versions page sets out stages from available to deprecated and retired and promises a migration guide for a major version, without a notice period (12 of 20). A sub-processor list updated 28 July 2026 and a per-Region availability list, and the August release note says queries stay in their Region (18 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Terms of service: read, states 6 of the 7 things a reader expects, and has 3 clauses that cost points (3.1 of 10)

## 7. Maintenance & community, 80 out of 100, up to 1.8 more on the total

Why it scored 80: Read for a closed service. The newest change is the expansion to four Regions on 15 October 2026 (30). Three dated changes in the 90 days, connector 1.2.0 on 20 July, filters and two Regions announced on 19 August, and four Regions on 15 October (20). Public release notes with RSS, What's New, re:Post and AWS Support. Reply times were not checked (10 of 15). Set-up is in the current AWS SDKs and the AgentCore CLI, and the call itself is plain MCP (15). Nothing package-specific to measure for a hosted connector (5 of 10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- `firstReleased` is 15 June 2026, the release date the connector versions page gives 1.1.0. The general availability post is dated 16 June 2026. Version 1.1.0's changes imply an earlier 1.0.0, which the page doesn't list or date.
- The Web Search page says improvements reach agents with no version upgrades, while the connector versions page says a target must be updated to adopt a new minor version. Which applies to index and snippet changes was not established.
- What happens when the 10 requests a second Web Search quota is exceeded, which error and whether `retryAfter` is returned. Only owner-set gateway rate limits are documented.
- Whether failed, empty or throttled searches are billed at the per-query rate.
- How long queries and results are kept, and whether they appear in gateway logs. No retention statement for Web Search was found.
- unchecked: the `tools/list` output, the tool's description text and any MCP annotations, since a call needs an AWS account. Only the published input schema was read.
- unchecked: the AWS Customer Agreement and Data Processing Addendum were not read this run.
- unchecked: whether Web Search is in SOC scope itself. The AgentCore FAQ lists SOC and ISO programmes for AgentCore as a whole.

## Weaknesses

- No error responses for the tool are documented. The only response example has `isError` false
- The default connector version is 1.1.0, so request-level domain and date filters work only on a target pinned to 1.2.0
- Available in seven Regions, four of them added on 15 October 2026
- Results are third-party page text, and the Web Search pages give no prompt-injection guidance. Gateway guardrail policies are documented separately
- No retention period for queries was found. The docs say queries are served inside AWS and not sent to a third-party search engine

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Call `tools/list` on the gateway and take the tool's name from it. AWS's call examples name gateway tools by target and tool
- Keep `query` to 200 characters and `maxResults` between 1 and 25 (default 10). The published schema states these limits only in descriptions
- Check the target's connector version before sending `filters`. Version 1.1.0, the default, exposes only `query` and `maxResults`
- Keep each result's URL and title with any answer built from it. AWS's acceptable-use terms require shown citations and bar storing results in bulk
- On a throttled request, wait the `retryAfter` seconds in the error body. The default quota is 10 Web Search requests a second, adjustable

## What the review panel asked for

- State billing of failed and throttled calls
- Document the tool's error responses
- Express limits as schema keywords
- make 1.2.0 the default
- documented tool error responses
- injection guidance for results
- stated query retention

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.