feat(.rootcoz): add rootcoz analysis agents

[rnetser]

Summary

Add 7 domain-specific rootcoz subagents to .rootcoz/agents/. These agents are automatically loaded into rootcoz AI sessions during failure analysis and chat, providing specialized capabilities for classifying test failures.

Depends on: myk-org/rootcoz#184 — rootcoz currently does not expose the subagent tool, so these agents are discovered but cannot be invoked yet. Once that bug is fixed, these agents become functional.

Agents

Agent	Purpose	Tools
fixture-tracer	Trace pytest fixture chains via --setup-plan	bash
wait-param-auditor	Audit wait params/timeouts for CODE ISSUE elimination	bash
quarantine-checker	Check quarantine (xfail+QUARANTINED) and Jira conditional skips	bash
teardown-cascade-detector	Detect cross-test contamination from failed teardowns	bash
version-extractor	Extract environment versions for the mandatory Environment block	read-only
product-code-investigator	Trace failures into upstream Go source code (KubeVirt, CDI, HCO)	bash
marker-analyzer	Analyze pytest markers for hardware/env requirements	bash

How it works

1. rootcoz clones this repo during analysis/chat
2. copy_rootcoz_pi_resources() copies .rootcoz/agents/ → <workspace>/.pi/agents/
3. Pi's DefaultResourceLoader discovers them
4. The AI delegates to agents via the subagent tool for specialized analysis tasks

Design decisions

• All agents use OPENSHIFT_VIRTUALIZATION_TEST_IMAGES_ARCH=amd64 to bypass cluster dependency during pytest introspection
• All agents have pytest fallback instructions (grep/read static analysis when pytest fails)
• version-extractor is the only read-only agent (no bash) — it only parses JSON files
• Agents use standardized input placeholders (<test_node_id>, <test_file_path>, etc.)
• Hybrid output format: Summary + Details (agent-specific) + Classification Impact
• No model override — agents inherit from the parent rootcoz session
• Trusted-repo security warning included on agents that run pytest

Review notes

• These agents were reviewed by 6 parallel reviewers (5 subagents + Cursor peer review) and iterated through fix cycles
• The quarantine-checker correctly documents both mechanisms: automation quarantine (xfail + QUARANTINED) and Jira conditional skip (pytest.mark.jira + run=False)
• Console output patterns in teardown-cascade-detector match this repo's custom pytest_report_teststatus format

What this PR does / why we need it:

Adds 7 rootcoz subagents for specialized failure analysis during rootcoz AI sessions.

Which issue(s) this PR fixes:

Special notes for reviewer:

Depends on myk-org/rootcoz#184 for the subagent tool to be exposed.

jira-ticket:

Co-authored-by: Claude noreply@anthropic.com

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 35df7493-ffbe-42b5-944b-bc17934bc0d1

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/rootcoz-agents

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}