Execution Risk Analyzer

Prerequisites

Required MCP Servers:

• aap-mcp-job-management - Job template lookup and launch parameter inspection
• aap-mcp-inventory-management - Inventory and host data

Verification: Run the aap-mcp-validator skill with these 2 servers before proceeding.

When to Use This Skill

Use this skill when:

• User requests a job execution targeting any environment
• Before any job template launch (as part of governance-executor workflow)
• User asks to check if an execution is safe
• User asks to validate execution parameters

Do NOT use when:

• Actually launching the job (use governed-job-launcher skill after this skill)
• Troubleshooting a failed job (use job-failure-analyzer skill)
• Assessing platform governance (use governance-readiness-assessor skill)

Workflow

Step 1: Consult Execution Governance Documentation

CRITICAL: Document consultation MUST happen BEFORE any MCP tool invocations.

Document Consultation (REQUIRED - Execute FIRST):

1. Action: Read execution-governance.md using the Read tool to understand inventory risk classification, extra_vars safety scanning, and governance controls
2. Output to user: "I consulted execution-governance.md which cites Red Hat's Security Best Practices and Job Templates documentation for execution governance controls."

Step 2: Identify the Job Template

MCP Tool: job_templates_list (from aap-mcp-job-management) Parameters:

• search: "<template_name_from_user_request>"
• page_size: 10

If the user provides a template ID directly:

MCP Tool: job_templates_retrieve (from aap-mcp-job-management) Parameters:

• id: "<template_id>"

Step 3: Inspect Launch Parameters

MCP Tool: job_templates_launch_retrieve (from aap-mcp-job-management) Parameters:

• id: "<template_id>"

This returns the template's expected extra_vars, defaults, required fields, and inventory configuration.

Step 4: Classify Inventory Risk

MCP Tool: inventories_list (from aap-mcp-inventory-management) Parameters:

• page_size: 100

Identify the target inventory from the job template configuration or user-provided override. Apply the risk classification from execution-governance.md:

For unclassifiable inventories, check host count:

MCP Tool: hosts_list (from aap-mcp-inventory-management) Parameters:

• search: "<inventory_name>"
• page_size: 1

Transparency note: Per execution-governance.md, inventory risk classification is this agent's implementation of Red Hat's recommendation to "use separate inventories for production and development environments" (Controller Best Practices).

Step 5: Pre-Execution Context Analysis (Scenario-Driven)

Per the "Pre-Execution Context Analysis" section of execution-governance.md, examine the template's operational context. These queries adapt the risk assessment to the specific scenario.

5a. Job History (always check):

MCP Tool: jobs_list (from aap-mcp-job-management) Parameters:

• job_template: "<template_id>"
• order_by: "-finished"
• page_size: 5

Examine the most recent 5 runs. Report:

• If all successful → "Clear history"
• If most recent failed → WARN: "Last run of this template failed. Investigate before re-executing."
• If 2+ consecutive failures → ELEVATED: "Template has failed [N] consecutive times."
• If 0 runs → INFO: "First execution of this template -- extra caution recommended."

5b. Launch Configuration (always check -- data available from Step 3):

From the job_templates_launch_retrieve response (Step 3), examine:

• ask_job_type_on_launch: If false AND risk is CRITICAL/HIGH → Warn that check mode cannot be enforced at launch time
• ask_limit_on_launch: If false AND risk is CRITICAL → Warn that phased rollout is not possible
• ask_diff_mode_on_launch: If false → Note that diff mode cannot be enabled at launch

5c. Notification Bindings (check for CRITICAL/HIGH risk):

MCP Tool: job_templates_retrieve (from aap-mcp-job-management) Parameters:

• id: "<template_id>"

Check notification association fields. If no failure notifications are bound: Report: "Per Red Hat's Notifications documentation (Ch. 25), this template has no failure notification. If execution fails, no one will be automatically alerted."

5d. Workflow Coverage (check for CRITICAL/HIGH risk):

MCP Tool: workflow_job_templates_list (from aap-mcp-job-management) Parameters:

• page_size: 100

If no workflows reference this template: Report: "Per Red Hat's Workflow documentation (Ch. 9), this template runs standalone -- not wrapped in a workflow with approval gates or failure paths."

5e. Module Analysis (check if job history exists):

If Step 5a found previous runs, examine events from the most recent job:

MCP Tool: jobs_job_events_list (from aap-mcp-job-management) Parameters:

• id: "<most_recent_job_id>"
• page_size: 50

If events show ansible.builtin.shell or ansible.builtin.command usage: Report specific warning: "This playbook uses shell/command modules ([X] tasks). Check mode will SKIP these tasks entirely."

5f. Adaptive Risk Adjustment:

Based on collected signals, adjust the base risk level per execution-governance.md's "Adaptive Risk Enhancement" table:

• HIGH + never run + check mode not overridable → elevate to CRITICAL
• LOW + recent failures → elevate to MEDIUM
• Any risk + no notifications on production target → add advisory flag

Step 6: Scan Extra Vars for Secrets

Inspect the extra_vars that would be passed to the job. Check both:

1. Default extra_vars from the template (from Step 3)
2. User-provided extra_vars overrides

Secret detection (per execution-governance.md):

• Key names containing (case-insensitive): password, secret, token, api_key, apikey, private_key, ssh_key, access_key, auth
• Values that look like tokens: long alphanumeric strings, base64, prefixes like sk-, ghp_, Bearer

Transparency note: Per execution-governance.md, secret scanning implements Red Hat's recommendation to "Remove user access to credentials" (Ch. 15, Sec. 15.1.4) by detecting plain-text secrets in extra_vars that should be managed via AAP credentials.

Step 7: Generate Risk Report

Output format:

## Execution Risk Analysis

**Job Template**: [name] (ID: [id])
**Target Inventory**: [name]
**Base Risk Level**: [CRITICAL / HIGH / MEDIUM / LOW]
**Adjusted Risk Level**: [same or elevated, with rationale if different]

### Risk Classification

Per Red Hat's *Controller Best Practices*: "Use separate inventories for production and development environments."

**Inventory signal**: [name pattern match or host count]
**Risk level**: [CRITICAL/HIGH/MEDIUM/LOW]
**Governance required**: [check mode + approval / confirmation only / direct execution]

### Operational Context [section only appears when signals found]

| Signal | Finding | Impact |
|---|---|---|
| Job History | [clear / last run failed / N consecutive failures / first run] | [proceed / investigate / elevated caution] |
| Check Mode Override | [available / not overridable] | [governance enforceable / template modification needed] |
| Phased Rollout | [available / not overridable] | [limit parameter works / all-or-nothing execution] |
| Notifications | [failure notifications bound / none configured] | [team will be alerted / silent failures] |
| Workflow Coverage | [wrapped in workflow / standalone] | [approval gates available / no governance gates] |
| Playbook Modules | [standard modules / shell/command detected ([X] tasks)] | [check mode reliable / partial dry-run coverage] |

### Extra Vars Safety Scan

Per Red Hat's *Security Best Practices* (Ch. 15, Sec. 15.1.4): "Remove user access to credentials."

| Check | Status | Detail |
|---|---|---|
| Secret-like key names | [PASS/FAIL] | [findings] |
| Plain-text values | [PASS/FAIL] | [findings] |

### Recommendation

[Based on risk level AND operational context, what governance controls should be applied]
[If risk was elevated: explain why and cite the triggering signals]

If secrets are found: BLOCK the execution and recommend using AAP credentials instead.

If CRITICAL/HIGH risk: Recommend check mode execution before full run.

If risk was elevated by operational signals: Explain the elevation clearly (e.g., "Risk elevated from HIGH to CRITICAL because this template has never been run and check mode cannot be overridden").

If LOW risk with no signals: Approve for direct execution with user confirmation.

Dependencies

Required MCP Servers

• aap-mcp-job-management - Job template data
• aap-mcp-inventory-management - Inventory and host data

Required MCP Tools

• job_templates_list (from job-management) - Find template by name
• job_templates_retrieve (from job-management) - Get template details and notification bindings
• job_templates_launch_retrieve (from job-management) - Get launch parameters and prompt-on-launch config
• jobs_list (from job-management) - Check recent job history for the template
• jobs_job_events_list (from job-management) - Analyze previous run's module usage
• workflow_job_templates_list (from job-management) - Check workflow coverage
• inventories_list (from inventory-management) - List inventories
• hosts_list (from inventory-management) - Host count for unclassifiable inventories

Related Skills

• aap-mcp-validator - Prerequisite validation
• governed-job-launcher - Next step after risk analysis passes
• execution-summary - Audit trail

Reference Documentation

• execution-governance.md - Risk classification and safety scanning reference

Example Usage

User: "Execute the security patch on production"

Agent:

1. Reads execution-governance.md
2. Reports: "I consulted execution-governance.md which cites Red Hat's Security Best Practices and Controller Best Practices."
3. Finds "Security Patch" template via job_templates_list
4. Inspects launch parameters via job_templates_launch_retrieve
5. Identifies "Production" inventory → CRITICAL base risk
6. Adapts: Checks job history → last run failed → adds WARN signal
7. Adapts: Checks launch config → ask_job_type_on_launch: true → check mode enforceable
8. Adapts: Checks notifications → no failure notifications → flags gap
9. Adapts: Checks previous run events → finds shell/command tasks → elevates check mode warning
10. Scans extra_vars → no secrets found
11. Reports: "Risk Level: CRITICAL (confirmed). Operational context: last run failed, no failure notifications configured, playbook uses 2 shell tasks that check mode will skip. Recommend investigating last failure before proceeding."