Skip to content

Day2/workshop exercises completed#11

Open
Gausen78 wants to merge 3 commits into
victorsteeb:mainfrom
Gausen78:day2/workshop-exercises-completed
Open

Day2/workshop exercises completed#11
Gausen78 wants to merge 3 commits into
victorsteeb:mainfrom
Gausen78:day2/workshop-exercises-completed

Conversation

@Gausen78

Copy link
Copy Markdown

No description provided.

Gausen78 and others added 3 commits June 22, 2026 16:10
…ng, and specialist gaps

- Coordinator now spawns one specialist per issue category instead of forcing a single specialist
  per ticket; multi-issue tickets no longer silently drop one problem
- Added ESCALATION TRIGGERS section so security requests (pentest auth, vuln disclosures) are
  routed to escalate_to_human and closed as "escalated" rather than falsely resolved
- Added change_plan tool to billing specialist (could diagnose plan issues but not act on them)
- Documented since_hours param on read_error_logs in technical specialist prompt
- Removed stale TODO comment from account tools that pointed to a non-existent tool
- Updated client brief template with full diagnostic arc including regression and escalation fix

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
- Building_an_Eval.ipynb: completed eval suite for Meridian coordinator
  with eval results across 12 runs stored in eval_results/
- Inference_Optimization.ipynb: completed ClauseScan v1 optimization sprint
  achieving 39× cost reduction ($0.1639 → $0.0042/contract) and 11×
  latency improvement (35.7s → 3.1s p50) with 100% accuracy held on
  both training and holdout sets
- Added summary.md files for both exercises documenting findings,
  SLA analysis, and lessons learned

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Always-on agent that processes customer campaign emails through Claude
Haiku for sentiment analysis, streams results to a FastAPI dashboard via
SSE, fires marketing team alerts on sentiment degradation, and includes
a 40-case eval notebook with LLM-as-judge scoring via Sonnet.

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant