# Vexlo

- [Introduction](https://docs.vexlo.ai/index.md): What Vexlo records, the three processes that get it from your code to a reader, and where to go next.
- [Quickstart](https://docs.vexlo.ai/quickstart.md): Run the server, the demo service, and the relay daemon on one machine, then read a trace back over the /v1 API.
- [The problem](https://docs.vexlo.ai/concepts/the-problem.md): Recorded signals are chosen in advance, which is why they run out at incident time for people and for AI agents alike.
- [Logs, spans, and execution graphs](https://docs.vexlo.ai/concepts/logs-spans-execution-graphs.md): A row-by-row comparison of what each one records, and why logging every function's values is not the same thing.
- [The execution graph](https://docs.vexlo.ai/concepts/execution-graph.md): The primitive, the three words for its parts, and one real trace read down to its captured values.
- [Use cases](https://docs.vexlo.ai/concepts/use-cases.md): Six questions execution data answers, and which part of the data answers each one.
- [Install](https://docs.vexlo.ai/sdk/install.md): Get the package into a process: requirements, the install from a checkout, and the import check.
- [Initialise and record a first trace](https://docs.vexlo.ai/sdk/initialise.md): Call init() once at startup, mark two functions, and read the records the SDK wrote to the spool directory.
- [Mark functions](https://docs.vexlo.ai/sdk/mark-functions.md): The two markers, @vexlo.trace and @vexlo.step, what each records, how retries group into attempts, and what the SDK never does to your function.
- [Tags and cross-service correlation](https://docs.vexlo.ai/sdk/tags.md): Declare tag keys at init(), add tags from inside the root, and relate traces across services by a shared key.
- [What gets captured](https://docs.vexlo.ai/sdk/what-gets-captured.md): Which slots an attempt carries, how each Python type is rendered, the bounds on a value, redaction, the error slot, and value sampling.
- [Context across threads](https://docs.vexlo.ai/sdk/context-across-threads.md): Where the current trace lives, why a thread pool loses it, and how carrier() and attach() carry it across.
- [Configuration reference](https://docs.vexlo.ai/sdk/configuration.md): Every init() argument and environment variable, what raises, the capture policy file, flush() and shutdown, and the in-memory bounds.
- [Install and run the server](https://docs.vexlo.ai/server/install.md): Build the vexlo binary from a checkout, start the Vexlo server, and check its configuration and retention arithmetic before it takes traffic.
- [Authentication and tokens](https://docs.vexlo.ai/server/authentication.md): Mint, list, and revoke API tokens, read a token's scope back from the server, and run the two modes that replace tokens: no-auth and header-trust.
- [The relay daemon](https://docs.vexlo.ai/server/relay-daemon.md): Run vexlo relay run on each host: what it reads from the spool directory, what it sends to the server, and how to watch and reload it.
- [Capture policy](https://docs.vexlo.ai/server/capture-policy.md): Lower what a service's SDK captures from the server side, per service or team-wide, without redeploying the service.
- [Retention](https://docs.vexlo.ai/server/retention.md): Shape is bounded by a day count and values by a byte budget. From about 1.24 traces per second the default 20 GiB budget runs dry before the 7-day shape count, and a trace keeps its shape long after the bytes behind its slots are reclaimed.
- [Conventions](https://docs.vexlo.ai/api/conventions.md): The headers, filters, fields, error shape, pagination, and rate limits every /v1 route shares.
- [Traces](https://docs.vexlo.ai/api/traces.md): The four trace routes: search the list, read one trace's shape, fetch its values, find traces that share a tag.
- [Functions](https://docs.vexlo.ai/api/functions.md): The function index: which marked functions ran, who called them, and example calls with captured values.
- [Tags, coverage, health](https://docs.vexlo.ai/api/tags-coverage-health.md): The tag keys a team has declared, per-service coverage and loss counters, and the server's own health.
- [Ingest](https://docs.vexlo.ai/api/ingest.md): The three routes an ingest token opens: post record batches, send heartbeats, set a capture policy.
- [Identity and bootstrap](https://docs.vexlo.ai/api/identity.md): What a credential resolves to, what the server enforces, and the two documents a client can read before it holds a token.
- [Install and configure](https://docs.vexlo.ai/mcp/install.md): Build vexlo-mcp from a checkout, point a client at it, and verify the handshake against a running server.
- [Startup checks and exit codes](https://docs.vexlo.ai/mcp/startup-and-exit-codes.md): What the process checks before it registers a tool, where its diagnostics go, and what each exit code means.
- [Find](https://docs.vexlo.ai/mcp/tools/find.md): The seven tools that name what exists, with their arguments, bounds, and captured responses.
- [Shape](https://docs.vexlo.ai/mcp/tools/shape.md): The two tools that return one trace's full step tree or one function's observed edges, with previews instead of value bytes.
- [Values](https://docs.vexlo.ai/mcp/tools/values.md): The two tools that return captured bytes: one trace's slots by path, and up to five worked examples of a function.
- [Debugging with an agent](https://docs.vexlo.ai/mcp/debugging-with-an-agent.md): Four tool calls from a symptom to the captured value that explains it, with what each one cost.
- [Errors, rate limits, retries](https://docs.vexlo.ai/mcp/errors-and-rate-limits.md): The two shapes a failed tool call takes, the nine error kinds and what each one means for an agent, and the one retry the process makes by itself.
- [Token costs](https://docs.vexlo.ai/mcp/token-costs.md): What one call of each tool costs in bytes and tokens, how it was measured, and which bounds those numbers set.
- [Troubleshooting](https://docs.vexlo.ai/troubleshooting.md): The ways capture stops without saying so, the signal that separates each one from its neighbours, and the command that produces that signal.
- [Glossary](https://docs.vexlo.ai/glossary.md): The term this site uses for each concept, with the synonyms it does not use.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.