feat(cloudformation): Add DevOps Agent alarm investigations template

[LorenzoBoccaccia]

Description

Add a CloudFormation addon that forwards a single CloudWatch alarm to a DevOps Agent generic (HMAC) webhook to open an investigation, plus a combined cloudformation/ README documenting it alongside the existing skill-policies template.

The signing Lambda hardens the forwarding path: layered SSRF egress checks with connect-to-validated-IP pinning, HMAC-SHA256 request signing, alarm-ARN match to reject spoofed/mismatched events, retry-stable incidentId for dedup, staleness cutoff matching the retry window, and least-privilege IAM. Reserved concurrency and log retention are configurable; the webhook HMAC key is read from a Secrets Manager secret (SecretString or SecretBinary).

Type of change

• New skill
• New custom agent
• Update to an existing skill or agent
• Documentation or infrastructure change

Testing

This change is a CloudFormation automation add-on (not a skill), so Agent Skill
Eval does not apply. It was validated at four levels — template, unit, live AWS,
and a real DevOps Agent webhook.

1. Template validation

• aws cloudformation validate-template — passed (6 parameters recognized,
  CAPABILITY_IAM).
• Full YAML parse with CloudFormation intrinsic tags — well-formed.

2. Local unit / adversarial tests of the inline signing Lambda (python3.12, the
target runtime) — 24/24 passing, covering:

• SSRF egress classifier: blocks RFC1918, loopback, link-local, 169.254.169.254
  (IMDS), CGNAT 100.64/10, TEST-NET/reserved, and IPv4-embedded-in-IPv6
  (::ffff:, NAT64 64:ff9b::, 6to4); allows public v4/v6.
• Log-injection: control chars / newlines / NUL stripped and ARN capped at 2048.
• Secret parsing: raw string, JSON key fields, and UTF-8 SecretBinary; rejects
  empty / missing-field / non-string.
• Staleness cutoff (8h) and ALARM-state gating.
• Alarm-ARN match: mismatched / empty / missing resources[0] rejected and logged.
• Retry-stable incidentId (event id, then alarmArn+state-timestamp; never
  wall-clock).
• HMAC signature: base64(HMAC-SHA256(key, "timestamp:payload")).
• End-to-end against a local TLS mock that recomputed the HMAC and validated the
  JSON schema through the real _post / pinned-TLS connect.

3. Live AWS deploy-and-teardown (throwaway dev account, all resources tagged,
fully torn down afterwards):

• Stack reached CREATE_COMPLETE (IAM role, inline-packaged Lambda, EventBridge
  rule, log group, conditional reserved concurrency).
• Invoked the deployed Lambda with a synthetic ALARM event: exercised Secrets
  Manager fetch → HMAC sign → SSRF resolve → TLS POST and handled the endpoint
  response.
• Spoofed event (wrong alarm ARN) → skipped: alarm ARN mismatch.
• Confirmed DeletionPolicy: Delete behavior; stack, log group, and secret removed.

4. Real DevOps Agent webhook contract (manual) — posted one signed request to a
live generic (HMAC) webhook: HTTP 200, confirming the signing scheme
(x-amzn-event-timestamp / x-amzn-event-signature) and the payload schema
(eventType/incidentId/action/priority/title/description/timestamp/
data) are accepted as-is.

License confirmation

• By submitting this pull request, I confirm that my contribution is made under the terms of the Apache License 2.0.

[yprikhodko]

Minor: AgentName parameter: rename + make optional

AgentName is required but purely cosmetic — it only sets the DevOpsAgent= tag and, per its own description, "does not affect routing." Two small issues:

Misleading name. "Agent name" reads as if it selects/routes to a specific agent, so a deployer may think it must match their real agent or that a wrong value breaks delivery. Suggest AgentNameTag to signal it's metadata.
Needlessly required. Forcing a value that has no functional effect adds friction.
Proposed:

Rename AgentName → AgentNameTag (param, the four !Ref tag blocks, and the ParameterLabels/ParameterGroups entries).
Make it optional: Default: !Ref AWS::StackName, drop MinLength: 1.
Trim description to: "Value applied as the DevOpsAgent tag on created resources (identification / cost allocation only; does not affect routing)."