# HotTea for Hatchways

HotTea adds standardized AI-agent access and reviewer evidence to repo-based assessments without proctoring, locked browsers, or fake editors.

## Fit

- Preserves GitHub-style practical tasks, automated tests, hidden grading, and human review.
- Adds a provisioned developer VM with Codex/Claude recorder wrappers.
- Packages Claude JSONL exports, terminal evidence, git snapshots, PR metadata, test output, hidden-test results, and an ATS-ready reviewer note.
- Keeps GitHub Actions as the default integration path while the optional GitHub App install surface remains dormant until configured.

## Live Proof

- Proof page: https://hottea.ai/hatchways
- Sample reviewer packet: https://hottea.ai/sample-report
- Sample reviewer Markdown: https://hottea.ai/sample-report.md
- Sample evidence JSON: https://hottea.ai/api/sample-evidence
- GitHub Action adapter: https://hottea.ai/github-action.yml
- Integration guide: https://hottea.ai/hatchways/integration.md
- Reviewer calibration: https://hottea.ai/hatchways/calibration.md
- Pilot kit: https://hottea.ai/hatchways/pilot.md
- OpenAPI: https://hottea.ai/openapi.yaml

## Sample Packet

- Assessment: Fix invite acceptance regression
- Candidate role: Senior Backend Developer
- Reviewer aid score: 9/10
- Suggested score: undefined/10
- Hidden tests: 3/3 passed
- AI session transcripts: 1
- Claude JSONL exports: 1
- Git snapshots: 2
- Test runs: 1
- VM events: 6

## Integration

1. Add the workflow from /github-action.yml to a candidate repository.
2. Create an HotTea candidate session and provide the signed applicant workspace URL with the Hatchways task.
3. Let the candidate work in the developer VM or attach outside AI evidence when required.
4. Review the HotTea packet beside Hatchways' existing grading and hidden-test output.

## Current Blockers

- The real GitHub App still requires account-level creation and GITHUB_APP_INSTALL_URL deployment configuration.
- Until that app exists, GitHub Actions remains the verified fallback adapter.

## Reviewer Note

```
HotTea reviewer note
Candidate: Alex Chen
Assessment: Fix invite acceptance regression
Score: 9/10 (rubric) · Suggested: 9/10 (AI fluency + verification)
Reviewer scores: not yet scored
Evidence captured: 1 AI session transcript(s), 1 Claude JSONL export(s), 0 external AI evidence upload(s), 2 git snapshot(s), 1 test run(s), 6 recorder event(s)
Summary: Kept the production change small. Verified the valid pending path still passes after changing branch order.
Verification observed: observable
AI fluency observed: observable in recorded AI session evidence
Post-interview transcript review: no transcript artifact stored
Report: https://hottea.ai/report/sess_hatchways_sample?token=rpt_sample
Suggested follow-up questions:
1. Walk through the highest-impact change and why that was the right first move.
2. Explain one tradeoff or shortcut from the assessment and what would change in production.
3. Describe how AI was used, what was accepted or rejected, and how the result was verified.
```