scope-exploit-playbook

SCOPE Exploit Playbook

Use this skill after scope-exploit Gate 3 approval and before Gate 4 display.

The top-level agent owns permission discovery, path reasoning, research dispatch, gates, artifact writes, and results.json. This skill owns the playbook document shape only.

Inputs

• TARGET_ARN
• RUN_ID
• DISCOVERY_MODE: standalone or audit
• discovery summary
• operator-approved paths
• research results from scope-research
• AWS_CLI_REPLAY from scope-awscli-replay
• real account IDs, ARNs, resource names, and permission evidence

Hard Rules

• Generate-only. Do not execute AWS CLI commands from this skill.
• Do not deploy or mutate AWS resources. The playbook may contain operator-reviewed commands, but SCOPE agents must not run them.
• Use actual ARNs, account IDs, resource names, and permissions from discovery. No placeholders except operator-supplied values already marked in AWS_CLI_REPLAY.
• Use command blocks from AWS_CLI_REPLAY.paths[].commands[].
• Start each path with narrative context before commands.
• Order steps by technical dependency, not quietness or stealth.
• Persistence and post-exploitation sections require explicit operator approval before inclusion. If the operator approved only escalation paths, omit those sections or mark them skipped.
• Include a minimal IAM policy document per path.
• Do not include CloudTrail event names, GuardDuty finding types, detection likelihood, OPSEC notes, SOC recommendations, numeric confidence scores, or stealth-ordering headers.
• Do not tag steps with static CloudTrail visibility classes. Detection and visibility analysis belongs to scope-controls and scope-investigate.
• Do not invent permissions, resources, or prerequisites.

Required Format

# Red Team Playbook — [TARGET_ARN]

**Run ID:** [RUN_ID]
**Discovery mode:** [standalone | audit]
**Generated:** [ISO-8601 timestamp]

## Discovery Summary

[What was probed or loaded, what succeeded, what was denied, and what those results revealed.]

## Path 1: [Descriptive Name]

[Narrative explanation of the attack chain, the permissions that enable it, and where the chain leads. Weave research citations inline when present.]

### Step 1: [Action Description]

```bash
[Command from AWS_CLI_REPLAY.paths[].commands[]]

[Why this step comes first and what it enables.]

Establish Persistence

[Persistence steps available from this path.]

Post-Exploitation

[Data access, lateral movement, or impact unlocked by the path.]

IAM Policy Document

{
  "Version": "2012-10-17",
  "Statement": []
}