TraceLane

Agent run tracing — spans for LLM, tool, retrieval, and handoff steps with tokens, credits, and failure receipts.

Part of Talocode. Hosted at /v1/tracelane/*.

| | | |--|--| | npm | npm install @talocode/tracelane | | Cloud | https://api.talocode.site/v1/tracelane/* | | Auth | TALOCODE_API_KEY |

Why TraceLane?

Production agents fail silently. You need a receipt chain: which tool ran, what failed, cost, and duration — not only the final answer.

Install

npm install @talocode/tracelane

Local engine

import { createRun, startSpan, endSpan, completeRun } from '@talocode/tracelane'

const run = createRun({ name: 'support-agent' })
const span = startSpan(run.id, { name: 'search', kind: 'tool' })
endSpan(run.id, span.id, { status: 'ok', costCredits: 5 })
const done = completeRun(run.id)

Cloud client

import { TraceLaneClient } from '@talocode/tracelane'
const t = new TraceLaneClient({ apiKey: process.env.TALOCODE_API_KEY })
const run = await t.createRun({ name: 'job' })

Routes / credits

| Action | Credits | |--------|---------| | Create run | 1 | | Start/end span | 0 | | Complete run | 1 | | List / get | 0–1 | | Export | 2 |

v0.1 store is process-local (in-memory).

Talocode ecosystem

| Project | What it is | |---------|------------| | VerifyLane | Deterministic verification for agent code | | TraceLane | Agent run / span tracing | | HandoffLane | Schema handoff validation | | StyleLane | Repo constitution / taste packs | | SearchLane | Agent web search & research | | Tera | AI chat / writing / coding API | | Codra | Coding agent | | StackLane | Cloud keys, credits, spend caps | | GateLane | MCP gateway | | Agent Browser | Browser automation |

More: github.com/talocode · talocode.site · docs.talocode.site

License

MIT © Talocode