I will build a RAG and AI agent reliability regression pack in 48 hours
I will build a deterministic reliability regression pack for one authorized RAG or AI-agent workflow in 48 hours.
Included scope:
- 20–30 focused regression cases adapted to your supplied sanitized workflow
- Checks for missing or stale context, unsupported claims, and citation grounding
- Checks for malformed or failed tool output, timeout and retry-budget behavior
- Duplicate-action and idempotency-key checks where applicable
- Executable pytest or CLI harness with exact local run commands
- Prioritized findings and a concise handoff
Acceptance criteria:
- The agreed cases run deterministically from a clean documented setup
- Each failure identifies the relevant reliability class and evidence
- No production credentials, customer data, or private benchmark answers are required
- The delivered harness and documentation match the agreed sanitized contract
The fixed price covers one bounded workflow and one revision inside the agreed scope. Production deployment, live-service load testing, model fine-tuning, custody, and extraction of hidden evaluation data are outside scope and can be quoted separately.
Work starts only after LaborX escrow is funded. Please send only sanitized fixtures or a development contract.
Verified public sample (reusable demo, not prior client work):
https://github.com/bughunter-dotcom/rag-agent-reliability-sample/commit/1f419b3be3a1a74bf6c3db7d1c87ce0bbba92077
Fresh public clone result: 23/23 tests passed.
