Control what your coding agents spend on LLMs.
BurnLens is a proxy that sits between your coding agents and the Claude API, giving you budgets, caps and per-agent cost visibility.
Get early accessSilent budget drains
Coding agents run unattended and can burn through API budgets in hours. Teams find out when the invoice arrives.
When autonomous loops retry failed tasks, generate high-token context windows, or hit recursion loops, costs escalate without instant boundary enforcement.
How BurnLens works
Point base URL
Point your agent's API base URL at BurnLens instead of the default endpoint.
Set budgets & caps
Set budgets and caps per agent, project or team in plain configuration.
Real-time protection
See live spend and get blocked or alerted before overruns occur.
export ANTHROPIC_BASE_URL="https://proxy.burnlens.sairintechnology.com/v1"
export BURNLENS_AGENT_ID="claude-code-worker-01"
# 2. Output when an agent hits a budget threshold
[BurnLens] Request authenticated: agent=claude-code-worker-01 (Session spend: $14.82 / Budget: $15.00)
[BurnLens] HTTP 429: Session budget limit ($15.00) reached. Request halted before provider overrun.
Built on Claude
BurnLens is designed specifically to work seamlessly with the Claude API and Claude Code workflows.
Protocol-Level Compatibility
Proxies requests cleanly to Claude models (including Claude 3.5 Sonnet and Claude 3.7 Sonnet) without altering prompt payload schemas or streaming responses.
Claude Code Workflow Integration
Intercepts CLI and automated agent calls to measure cumulative input, cache write/read, and output token costs with provider-accurate pricing calculation.
Zero Latency Penalty
Transparent streaming passthrough ensures agents execute terminal edits, file modifications, and planning tasks with no perceived latency.
Get early access to BurnLens
Engineering teams running Claude Code or other coding agents.
Direct response from contact@sairintechnology.com. No spam.