CostEntry (observability/cost_attribution.py)

aiming-lab/AutoHarness · autoharness/observability/cost_attribution.py · python · MIT
ai

Cost attribution — fine-grained cost tracking per tool, agent, and workflow. Tracks token costs at multiple granularities: per-tool-call, per-agent, per-workflow, and per-session. Enables cost optimization and billing. Usage:: tracker = CostTracker() with tracker.track_call("Bash

# Source: https://github.com/aiming-lab/AutoHarness | License: MIT | Copyright 2026 AgentLint Contributors
# Provided via justclaw.it.com; keep this notice with the code.
"""Cost attribution — fine-grained cost tracking per tool, agent, and workflow.

Tracks token costs at multiple granularities: per-tool-call, per-agent,
per-workflow, and per-session. Enables cost optimization and billing.

Usage::

    tracker = CostTracker()

    with tracker.track_call("Bash", agent_id="explore-1") as call:
        # ... execute tool call ...
        call.record_tokens(input=1200, output=300)

    report = tracker.generate_report()
    print(report.by_tool)     # {"Bash": $0.012, "Read": $0.003, ...}
    print(report.by_agent)    # {"explore-1": $0.015, "plan-1": $0.008}
    print(report.total_cost)  #

... (truncated -- full source via MCP)

See the full source, get the GitHub permalink, and search for more like it.

Get a free API key