Skip to content
#

expected-cost

Star

Here are 2 public repositories matching this topic...

kapardhi03 / cost-aware-escalation-agent

Star 1

When should an AI agent stop replying and get a human? A cost-aware escalation policy: an LLM belief over hidden lead intent, scored against an asymmetric cost matrix, picking the lowest-expected-cost action. 100 cached cases, reproducible runs, and a preprint on where better calibration stops buying better decisions.

python reproducible-research calibration openai uncertainty-quantification isotonic-regression decision-theory human-in-the-loop escalation preprint conversational-ai belief-state llm-agents agent-evaluation expected-cost cost-sensitive-decisions
  • Updated Sep 18, 2026
  • Python

noqt / yumi-decision-ledger

Star 0

Hosted anonymous MCP/HTTP: PyPI wheel evidence; mathematical IPv4/IPv6 CIDR-set differences, counts and witnesses. Exact diagnostic planning for small deterministic models; offline decision-ledger, handoff, snapshot and opt-in command-evidence tools. Public non-secret inputs; not compatibility, reachability, routing or security validation.

python cli ipv6 pypi provenance diagnostics cidr python-wheels package-metadata release-engineering fault-isolation pep-425 mcp-server agent-skills agent-memory agent-handoff claude-skills decision-ledger expected-cost probe-planning
  • Updated Oct 25, 2026
  • Python

Add this topic to your repo

To associate your repository with the expected-cost topic, visit your repo's landing page and select "manage topics."

Learn more