Claude Skill

aws-vpc-dns-investigation

Use this skill when a name is not resolving as expected inside a VPC, or before applying a DNS control-plane change. Activate on symptoms such as NXDOMAIN or SERVFAIL from an EC2 instance, a hostname resolving to a public address when a private endpoint was expected, an AWS servi

LLM Mart · 0 points · 2 views 0 listing impressions 0 install-command copies
Download aws-tools-for-devops-agent-skills_aws-vpc-dns-investigation-d8f96e1.zip · 24 KB

Install

skills CLI npx skills add https://github.com/aws/tools-for-devops-agent/tree/main/skills/aws-vpc-dns-investigation
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install aws-tools-for-devops-agent@llmmart
Git git clone https://github.com/aws/tools-for-devops-agent.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole aws/tools-for-devops-agent collection as a plugin from our marketplace. Git is the plain clone.

README

AWS VPC DNS Investigation

This skill tells the AWS DevOps Agent when to reach for the aws-vpc-dns-diagnostics MCP server and in what order to use its tools, so a cold symptom like "this name will not resolve" leads to a consistent investigation instead of an ad hoc one.

⚠️ Important Notice

This skill is sample code, not intended for production use without additional review and testing. Users should validate in a non-production environment first. It is read-only: it drives observation and simulation tools and takes no action on your DNS configuration.

Purpose

The MCP server's bundled runbooks tell the agent how to interpret a DNS result once it has already decided to look. Nothing tells it when to look, or that the VPC attribute check has to come before any interpretation. This skill supplies that activation trigger and the investigation sequence.

The sequence matters because VPC DNS failures are ordered. enableDnsSupport gates the entire VPC resolver, so a resolution result read before that attribute is checked can be attributed to the wrong cause. Likewise, an answer from an unknown resolver is not evidence until you know which resolver answered, and agreement between resolvers is not the same as correctness.

Key Capabilities

  • Recognize VPC DNS symptoms from an operator's description without being told to use the skill
  • Enforce the precondition order: VPC attributes and host context before any interpretation of a resolution result
  • Drive live, multi-resolver comparison from inside the affected subnet, including resolver identity, rather than resolving from the agent's own vantage point
  • Route a proposed DNS change through symbolic simulation before it is recommended
  • Load the matching pattern runbook for a signature instead of reasoning from general knowledge
  • Report cross-account opaque constructs and missing SSM reachability as boundaries rather than inferring past them

Prerequisites

  • The aws-vpc-dns-diagnostics MCP server registered in your Agent Space with its tools allowlisted. The server and its deployment instructions are in this repository at mcp/aws-vpc-dns-diagnostics-mcp/
  • The scoped IAM roles deployed in each target account, and those account IDs supplied in the server's AllowedAccounts parameter. There is no default and a wildcard is refused
  • For the Mode A live probe tools, the target EC2 instance reachable through SSM: interface endpoints for ssm, ssmmessages, and ec2messages, and an instance role with AmazonSSMManagedInstanceCore
  • No additional DevOps Agent role permissions. The agent calls the MCP server, which assumes its own scoped roles in the target accounts

Limitations

  • The skill is guidance only. Without the MCP server registered and allowlisted, none of the tools it references are callable
  • Mode A requires SSM reachability to the target instance. An instance that is not SSM-managed can be reasoned about from configuration but not probed
  • Cross-account constructs shared with the target account may be enumerable while their contents remain opaque. These are reported as unknown, not absent
  • Simulation is symbolic. It predicts the effect of a change on resolution and does not apply, stage, or validate the change against the live control plane

Agent Types

  • Chat tasks - conversational DNS diagnosis and pre-change validation
  • Incident RCA - automated investigation where a failure may have a DNS cause

Uploading to AWS DevOps Agent

Register the MCP server first. The skill references its tools by name, and trigger behavior cannot be validated until those tools are present in the Agent Space.

Option A: Import from GitHub (recommended)

If you have a GitHub connection configured in your Agent Space, import this skill directly from the repository. In the DevOps Agent web app, go to Settings → Add Skill → Import from repository, then point to the skills/aws-vpc-dns-investigation directory.

Option B: Upload as a zip file

  1. Zip the directory, including only allowed extensions:

    cd skills
    zip -r aws-vpc-dns-investigation.zip aws-vpc-dns-investigation/ -i '*.md' '*.txt' '*.json' '*.yaml' '*.yml' '*.xml' '*.csv' '*.tsv' '*.html' '*.htm' '*.png' '*.jpg' '*.jpeg' '*.gif' '*.svg' '*.webp' '*.pdf' -x '*/.claude/*' '*/scripts/*' '*/README.md' '*/.skilleval.yaml' '*/.skilleval.yml' '*/CHANGELOG.md' '*/evals/*'
    
  2. In the AWS DevOps Agent web app, go to the Skills page.

  3. Click Add skill → Upload skill.

  4. Drag and drop the zip file.

  5. Select the agent types: Chat tasks and Incident RCA.

  6. Click Upload.

Option C: Upload via the Asset API

Assign the skill to the CHAT and INCIDENT_RCA agent types. See Managing a skill end-to-end.

How to Use This Skill

Chat

  • "secretsmanager.us-east-1.amazonaws.com returns NXDOMAIN from this instance but works from another one."
  • "Our app is reaching a public IP for a service we put behind an interface endpoint."
  • "Would enabling private DNS on this VPC endpoint break anything?"
  • "What would a Resolver FORWARD rule for '.' pointing at on-prem do to the names that resolve today?"
  • "Show me the effective DNS configuration for this VPC, including anything inherited from a Route 53 Profile."

Investigation

  • "An application started failing after a network change last night. Resolution looks wrong from the subnet."
  • "Half our instances can reach the internal API by hostname and half cannot."
  • "A service endpoint stopped resolving after we associated a Route 53 Profile."
  • "IPv6 clients get a different answer than IPv4 clients for the same name."

Learn More

Skill manifest

Investigate VPC DNS Resolution

Use the tools on the connected aws-vpc-dns-diagnostics MCP server.

Step 1: Classify the request as Mode A or Mode B

Before calling any tool, determine which mode applies:

  • Mode A (live diagnosis): The operator reports a resolution symptom from a running instance. They provide an instance ID (or you can identify one). The goal is to observe what actually resolves and compare resolvers.
  • Mode B (pre-change validation): The operator asks whether a proposed DNS change is safe. They provide account, region, VPC, and a change descriptor. No instance is required.

If the request is ambiguous, ask the operator to clarify. Do not default to Mode A when the input lacks an instance ID, and do not default to Mode B when the operator describes a live symptom.

Step 2: Load safety rules

Regardless of mode, call get_sop with slug A-critical-safety-rules and follow every rule it contains. These are non-negotiable constraints on how you interpret results, handle opaque constructs, and report findings.


Mode A route: live diagnosis

Required inputs

account_id, region, instance_id, and the failing DNS name.

Tool sequence (in order)

  1. dns_probe_context — establishes VPC-attribute preconditions: enableDnsSupport, enableDnsHostnames, address family, DHCP option set. A resolution result means nothing until you know whether the VPC resolver is answering.
  2. dns_probe_compare — runs the allowlisted probe set inside the instance via SSM. Returns each resolver's answer and the resolver's own identity from hostname.bind. The VPC DHCP resolver is auto-added for comparison.
  3. get_sop — load the pattern runbook matching the observed signature (see trap-to-SOP mapping below).

Interpretation rules

  • If enableDnsSupport is false: load A-resolver-disabled-precondition. The VPC resolver is intentionally dark and every probe failure follows from that.
  • Compare the instance's /etc/resolv.conf (from the probe output) against the DHCP option set. A mismatch means the instance is not using the VPC-intended resolver.
  • Judge answers by name category (load A-name-category-classification), not by whether resolvers agree. Two resolvers returning the same wrong answer is still a failure.

Mode A trap-to-SOP mapping

Observed signature SOP slug
Custom resolver answers differently from VPC .2 A-custom-resolver-divergence
FORWARD rule and PHZ both match the name A-forward-vs-phz-precedence-collision
A record works, AAAA fails (or vice versa) A-address-family-divergence
enableDnsSupport is false A-resolver-disabled-precondition
General live comparison procedure A-mode-a-live-resolver-comparison

Reporting format for Mode A

Label every finding as Observed (ground truth from the probe). State which resolver answered and what it returned. When Mode A and Mode B produce different conclusions for the same name, Mode A wins because it is ground truth from inside the subnet.


Mode B route: pre-change validation

Required inputs

account_id, region, vpc_id, and a change descriptor (structured dict with type and type-specific fields). No instance required.

Tool sequence (in order)

  1. dns_simulate_effective_config — returns the VPC's effective DNS config: the union of directly attached resources and anything inherited through an associated Route 53 Profile, each construct tagged by source.
  2. dns_simulate_change — applies the proposed change symbolically and returns a per-name impact report (before/after, delta, traps, severity, volume).
  3. get_sop — load runbooks for any traps reported in the impact table (see trap-to-SOP mapping below).

Interpretation rules

  • Never recommend applying a change without simulating it first. A broad FORWARD rule, enabling private DNS on an interface endpoint, or a Profile association can silently redirect names that currently resolve correctly.
  • The candidate set is limited to API-derived names (PHZ records, rule domains, VPCE apexes, Firewall domain lists) or operator-supplied names. It is not exhaustive. State the coverage boundary.
  • If the operator supplies volumes (from Resolver Query Logs), names are ranked by traffic. This is enrichment; absence does not invalidate the simulation.

Mode B trap-to-SOP mapping

Trap label in impact report SOP slug
VPCE-shadow-NXDOMAIN B-vpce-shadow-nxdomain
broad-FORWARD-sweep B-broad-forward-sweep
flag-AND-mismatch B-flag-and-mismatch
DNS-Firewall-block B-dns-firewall-block
profile-union-shift B-profile-propagation-timing
General pre-change procedure B-mode-b-pre-change-validation

Reporting format for Mode B

Label every finding as Predicted (symbolic, not ground truth). State the candidate-set size, its source (API-derived or operator-supplied), and that names outside this set were not evaluated. Include the propagation timing caveat for Profile changes.


Cross-account opacity

Call get_sop with slug C-cross-account-opaque-constructs when the effective config or impact report contains opaque markers. Cross-account constructs shared via RAM or a Route 53 Profile may be enumerable but their contents are not readable from the consumer account. Report them as "present but unknown content" rather than treating them as absent or inferring past them.

Limitations

Call get_sop with slug C-limitations-and-boundaries and state the relevant boundaries to the operator. Key constraints:

  • All tools are read-only. Do not modify, delete, or create DNS resources.
  • Mode A requires SSM reachability (ssm, ssmmessages, ec2messages VPC endpoints and an instance role with AmazonSSMManagedInstanceCore).
  • Mode B candidate sets are not exhaustive. The "no impacts" conclusion applies only within the tested set.
  • Opaque constructs cannot be resolved from this account.
  • Resolver Query Log ingestion is not implemented; volumes must be supplied by the operator.

Final response requirements

Every response produced by this skill must include:

  1. Each finding labelled Observed (Mode A) or Predicted (Mode B).
  2. When both modes were used, state "Mode A wins" for any conflict.
  3. The candidate-set coverage: how many names, what source, what was not tested.
  4. Any opaque constructs and their impact on the conclusion.
  5. Recommended next steps or the specific change to apply (never apply it).

Prerequisites

Requires the aws-vpc-dns-diagnostics MCP server registered in the Agent Space with its tools allowlisted. The server is at mcp/aws-vpc-dns-diagnostics-mcp/. If the server is not registered or SSM is unreachable, report that as the blocker rather than guessing at the resolution path.

Files (tools-for-devops-agent)
  • evals
    • benchmark.json 54.9 KB
      Loading…
    • evals.json 4.6 KB
      Loading…
    • eval_queries.json 1.9 KB
      Loading…
    • report.json 1.6 KB
      Loading…
    • trigger_report.json 5.4 KB
      Loading…
  • .skilleval.yaml 77 B
    Loading…
  • CHANGELOG.md 232 B
    Loading…
  • README.md 6.4 KB
    # AWS VPC DNS Investigation
    
    This skill tells the AWS DevOps Agent when to reach for the
    [aws-vpc-dns-diagnostics MCP server](https://github.com/aws/tools-for-devops-agent/tree/main/mcp/aws-vpc-dns-diagnostics-mcp) and in
    what order to use its tools, so a cold symptom like "this name will not resolve"
    leads to a consistent investigation instead of an ad hoc one.
    
    ## ⚠️ Important Notice
    
    This skill is sample code, not intended for production use without additional
    review and testing. Users should validate in a non-production environment first.
    It is read-only: it drives observation and simulation tools and takes no action on
    your DNS configuration.
    
    ## Purpose
    
    The MCP server's bundled runbooks tell the agent how to interpret a DNS result once
    it has already decided to look. Nothing tells it when to look, or that the VPC
    attribute check has to come before any interpretation. This skill supplies that
    activation trigger and the investigation sequence.
    
    The sequence matters because VPC DNS failures are ordered. `enableDnsSupport`
    gates the entire VPC resolver, so a resolution result read before that attribute is
    checked can be attributed to the wrong cause. Likewise, an answer from an unknown
    resolver is not evidence until you know which resolver answered, and agreement
    between resolvers is not the same as correctness.
    
    ## Key Capabilities
    
    - Recognize VPC DNS symptoms from an operator's description without being told to
      use the skill
    - Enforce the precondition order: VPC attributes and host context before any
      interpretation of a resolution result
    - Drive live, multi-resolver comparison from inside the affected subnet, including
      resolver identity, rather than resolving from the agent's own vantage point
    - Route a proposed DNS change through symbolic simulation before it is recommended
    - Load the matching pattern runbook for a signature instead of reasoning from
      general knowledge
    - Report cross-account opaque constructs and missing SSM reachability as boundaries
      rather than inferring past them
    
    ## Prerequisites
    
    - The `aws-vpc-dns-diagnostics` MCP server registered in your Agent Space with its
      tools allowlisted. The server and its deployment instructions are in this
      repository at [`mcp/aws-vpc-dns-diagnostics-mcp/`](https://github.com/aws/tools-for-devops-agent/tree/main/mcp/aws-vpc-dns-diagnostics-mcp)
    - The scoped IAM roles deployed in each target account, and those account IDs
      supplied in the server's `AllowedAccounts` parameter. There is no default and a
      wildcard is refused
    - For the Mode A live probe tools, the target EC2 instance reachable through SSM:
      interface endpoints for `ssm`, `ssmmessages`, and `ec2messages`, and an instance
      role with `AmazonSSMManagedInstanceCore`
    - No additional DevOps Agent role permissions. The agent calls the MCP server,
      which assumes its own scoped roles in the target accounts
    
    ## Limitations
    
    - The skill is guidance only. Without the MCP server registered and allowlisted,
      none of the tools it references are callable
    - Mode A requires SSM reachability to the target instance. An instance that is not
      SSM-managed can be reasoned about from configuration but not probed
    - Cross-account constructs shared with the target account may be enumerable while
      their contents remain opaque. These are reported as unknown, not absent
    - Simulation is symbolic. It predicts the effect of a change on resolution and does
      not apply, stage, or validate the change against the live control plane
    
    ## Agent Types
    
    - **Chat tasks** - conversational DNS diagnosis and pre-change validation
    - **Incident RCA** - automated investigation where a failure may have a DNS cause
    
    ## Uploading to AWS DevOps Agent
    
    Register the MCP server first. The skill references its tools by name, and trigger
    behavior cannot be validated until those tools are present in the Agent Space.
    
    **Option A: Import from GitHub (recommended)**
    
    If you have a [GitHub connection configured](https://docs.aws.amazon.com/devopsagent/latest/userguide/connecting-to-cicd-pipelines-connecting-github.html)
    in your Agent Space, import this skill directly from the repository. In the DevOps
    Agent web app, go to Settings → Add Skill → Import from repository, then point to
    the `skills/aws-vpc-dns-investigation` directory.
    
    **Option B: Upload as a zip file**
    
    1. Zip the directory, including only allowed extensions:
    
       ```bash
       cd skills
       zip -r aws-vpc-dns-investigation.zip aws-vpc-dns-investigation/ -i '*.md' '*.txt' '*.json' '*.yaml' '*.yml' '*.xml' '*.csv' '*.tsv' '*.html' '*.htm' '*.png' '*.jpg' '*.jpeg' '*.gif' '*.svg' '*.webp' '*.pdf' -x '*/.claude/*' '*/scripts/*' '*/README.md' '*/.skilleval.yaml' '*/.skilleval.yml' '*/CHANGELOG.md' '*/evals/*'
       ```
    
    2. In the AWS DevOps Agent web app, go to the **Skills** page.
    3. Click **Add skill** → **Upload skill**.
    4. Drag and drop the zip file.
    5. Select the agent types: **Chat tasks** and **Incident RCA**.
    6. Click **Upload**.
    
    **Option C: Upload via the Asset API**
    
    Assign the skill to the `CHAT` and `INCIDENT_RCA` agent types. See
    [Managing a skill end-to-end](https://docs.aws.amazon.com/devopsagent/latest/userguide/about-aws-devops-agent-managing-assets.html#managing-a-skill-end-to-end).
    
    ## How to Use This Skill
    
    ### Chat
    
    - "secretsmanager.us-east-1.amazonaws.com returns NXDOMAIN from this instance but works from another one."
    - "Our app is reaching a public IP for a service we put behind an interface endpoint."
    - "Would enabling private DNS on this VPC endpoint break anything?"
    - "What would a Resolver FORWARD rule for '.' pointing at on-prem do to the names that resolve today?"
    - "Show me the effective DNS configuration for this VPC, including anything inherited from a Route 53 Profile."
    
    ### Investigation
    
    - "An application started failing after a network change last night. Resolution looks wrong from the subnet."
    - "Half our instances can reach the internal API by hostname and half cannot."
    - "A service endpoint stopped resolving after we associated a Route 53 Profile."
    - "IPv6 clients get a different answer than IPv4 clients for the same name."
    
    ## Learn More
    
    - [aws-vpc-dns-diagnostics MCP server](https://github.com/aws/tools-for-devops-agent/blob/main/mcp/aws-vpc-dns-diagnostics-mcp/README.md)
    - [Architecture](https://github.com/aws/tools-for-devops-agent/blob/main/mcp/aws-vpc-dns-diagnostics-mcp/docs/ARCHITECTURE.md)
    - [AWS DevOps Agent Skills documentation](https://docs.aws.amazon.com/devopsagent/latest/userguide/about-aws-devops-agent-devops-agent-skills.html)
    
  • SKILL.md 7.9 KB
    ---
    name: aws-vpc-dns-investigation
    description: Use this skill when a name is not resolving as expected inside a VPC, or before applying a DNS control-plane change. Activate on symptoms such as NXDOMAIN or SERVFAIL from an EC2 instance, a hostname resolving to a public address when a private endpoint was expected, an AWS service endpoint that stopped resolving after a VPC endpoint or Route 53 change, an application reaching the wrong IP, resolution that works from one instance but not another, IPv6 or dualstack resolution differences, a suspected on-premises forwarding or hybrid DNS problem, or a request to check whether enabling private DNS, adding a Resolver rule, associating a private hosted zone, attaching DNS Firewall, or associating a Route 53 Profile would break anything. It drives the aws-vpc-dns-diagnostics MCP server to observe live resolution from inside the subnet and to simulate a proposed change before it is applied.
    metadata:
      author: ddericco
      version: "1.0.1"
      aws-devops-agent-skills.agent-types: "CHAT, INCIDENT_RCA"
      aws-devops-agent-skills.aws-services: "Amazon VPC, Amazon Route 53, Amazon EC2, AWS Systems Manager"
      aws-devops-agent-skills.technical-domains: "Networking"
    ---
    
    # Investigate VPC DNS Resolution
    
    Use the tools on the connected `aws-vpc-dns-diagnostics` MCP server.
    
    ## Step 1: Classify the request as Mode A or Mode B
    
    Before calling any tool, determine which mode applies:
    
    - **Mode A (live diagnosis):** The operator reports a resolution symptom from a
      running instance. They provide an instance ID (or you can identify one). The
      goal is to observe what actually resolves and compare resolvers.
    - **Mode B (pre-change validation):** The operator asks whether a proposed DNS
      change is safe. They provide account, region, VPC, and a change descriptor.
      No instance is required.
    
    If the request is ambiguous, ask the operator to clarify. Do not default to
    Mode A when the input lacks an instance ID, and do not default to Mode B when
    the operator describes a live symptom.
    
    ## Step 2: Load safety rules
    
    Regardless of mode, call `get_sop` with slug `A-critical-safety-rules` and
    follow every rule it contains. These are non-negotiable constraints on how you
    interpret results, handle opaque constructs, and report findings.
    
    ---
    
    ## Mode A route: live diagnosis
    
    ### Required inputs
    
    account_id, region, instance_id, and the failing DNS name.
    
    ### Tool sequence (in order)
    
    1. `dns_probe_context` — establishes VPC-attribute preconditions: enableDnsSupport,
       enableDnsHostnames, address family, DHCP option set. A resolution result means
       nothing until you know whether the VPC resolver is answering.
    2. `dns_probe_compare` — runs the allowlisted probe set inside the instance via
       SSM. Returns each resolver's answer and the resolver's own identity from
       `hostname.bind`. The VPC DHCP resolver is auto-added for comparison.
    3. `get_sop` — load the pattern runbook matching the observed signature
       (see trap-to-SOP mapping below).
    
    ### Interpretation rules
    
    - If `enableDnsSupport` is false: load `A-resolver-disabled-precondition`. The
      VPC resolver is intentionally dark and every probe failure follows from that.
    - Compare the instance's `/etc/resolv.conf` (from the probe output) against the
      DHCP option set. A mismatch means the instance is not using the VPC-intended
      resolver.
    - Judge answers by name category (load `A-name-category-classification`), not by
      whether resolvers agree. Two resolvers returning the same wrong answer is still
      a failure.
    
    ### Mode A trap-to-SOP mapping
    
    | Observed signature | SOP slug |
    | --- | --- |
    | Custom resolver answers differently from VPC .2 | `A-custom-resolver-divergence` |
    | FORWARD rule and PHZ both match the name | `A-forward-vs-phz-precedence-collision` |
    | A record works, AAAA fails (or vice versa) | `A-address-family-divergence` |
    | enableDnsSupport is false | `A-resolver-disabled-precondition` |
    | General live comparison procedure | `A-mode-a-live-resolver-comparison` |
    
    ### Reporting format for Mode A
    
    Label every finding as **Observed** (ground truth from the probe). State which
    resolver answered and what it returned. When Mode A and Mode B produce different
    conclusions for the same name, **Mode A wins** because it is ground truth from
    inside the subnet.
    
    ---
    
    ## Mode B route: pre-change validation
    
    ### Required inputs
    
    account_id, region, vpc_id, and a change descriptor (structured dict with `type`
    and type-specific fields). No instance required.
    
    ### Tool sequence (in order)
    
    1. `dns_simulate_effective_config` — returns the VPC's effective DNS config: the
       union of directly attached resources and anything inherited through an
       associated Route 53 Profile, each construct tagged by source.
    2. `dns_simulate_change` — applies the proposed change symbolically and returns a
       per-name impact report (before/after, delta, traps, severity, volume).
    3. `get_sop` — load runbooks for any traps reported in the impact table
       (see trap-to-SOP mapping below).
    
    ### Interpretation rules
    
    - Never recommend applying a change without simulating it first. A broad FORWARD
      rule, enabling private DNS on an interface endpoint, or a Profile association
      can silently redirect names that currently resolve correctly.
    - The candidate set is limited to API-derived names (PHZ records, rule domains,
      VPCE apexes, Firewall domain lists) or operator-supplied names. It is not
      exhaustive. State the coverage boundary.
    - If the operator supplies `volumes` (from Resolver Query Logs), names are ranked
      by traffic. This is enrichment; absence does not invalidate the simulation.
    
    ### Mode B trap-to-SOP mapping
    
    | Trap label in impact report | SOP slug |
    | --- | --- |
    | VPCE-shadow-NXDOMAIN | `B-vpce-shadow-nxdomain` |
    | broad-FORWARD-sweep | `B-broad-forward-sweep` |
    | flag-AND-mismatch | `B-flag-and-mismatch` |
    | DNS-Firewall-block | `B-dns-firewall-block` |
    | profile-union-shift | `B-profile-propagation-timing` |
    | General pre-change procedure | `B-mode-b-pre-change-validation` |
    
    ### Reporting format for Mode B
    
    Label every finding as **Predicted** (symbolic, not ground truth). State the
    candidate-set size, its source (API-derived or operator-supplied), and that names
    outside this set were not evaluated. Include the propagation timing caveat for
    Profile changes.
    
    ---
    
    ## Cross-account opacity
    
    Call `get_sop` with slug `C-cross-account-opaque-constructs` when the effective
    config or impact report contains opaque markers. Cross-account constructs shared
    via RAM or a Route 53 Profile may be enumerable but their contents are not
    readable from the consumer account. Report them as "present but unknown content"
    rather than treating them as absent or inferring past them.
    
    ## Limitations
    
    Call `get_sop` with slug `C-limitations-and-boundaries` and state the relevant
    boundaries to the operator. Key constraints:
    
    - All tools are read-only. Do not modify, delete, or create DNS resources.
    - Mode A requires SSM reachability (ssm, ssmmessages, ec2messages VPC endpoints
      and an instance role with AmazonSSMManagedInstanceCore).
    - Mode B candidate sets are not exhaustive. The "no impacts" conclusion applies
      only within the tested set.
    - Opaque constructs cannot be resolved from this account.
    - Resolver Query Log ingestion is not implemented; volumes must be supplied by
      the operator.
    
    ## Final response requirements
    
    Every response produced by this skill must include:
    
    1. Each finding labelled **Observed** (Mode A) or **Predicted** (Mode B).
    2. When both modes were used, state "Mode A wins" for any conflict.
    3. The candidate-set coverage: how many names, what source, what was not tested.
    4. Any opaque constructs and their impact on the conclusion.
    5. Recommended next steps or the specific change to apply (never apply it).
    
    ## Prerequisites
    
    Requires the aws-vpc-dns-diagnostics MCP server registered in the Agent Space
    with its tools allowlisted. The server is at `mcp/aws-vpc-dns-diagnostics-mcp/`.
    If the server is not registered or SSM is unreachable, report that as the blocker
    rather than guessing at the resolution path.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.