Skip to content

feat: add wire decoders and proxy auth - #6

Open
moshloop wants to merge 90 commits into
mainfrom
feat/wire-decoders-proxy-auth
Open

moshloop wants to merge 90 commits into
mainfrom
feat/wire-decoders-proxy-auth

Conversation

@moshloop

@moshloop moshloop commented Nov 3, 2026

Copy link
Copy Markdown
Member

What

  • Add passive metadata decoders for AWS, Git, and SFTP wire traffic
  • Add per-endpoint proxy authorization with token scope, endpoint allow-lists, expiry, and revocation checks
  • Bump flanksource/deps action to v1.0.50 in CI and release workflows

Why

  • Decoders record request metadata while leaving signed requests and file or pack contents unmodified
  • Proxy auth checks every request and accepts Basic or Bearer credentials

Notes

  • Includes unit tests for the new decoders and proxy auth

moshloop and others added 30 commits November 3, 2026 05:33
…ckages

Trace search and filtering logic and the OTel search UI are now provided by the shared commons-db query/tracequery package and the clicky-ui data components. The local tracequery and otel webapp modules become thin adapters that delegate to them, so behavior is unchanged while the implementation lives in one place. Also adds a proxy token scope, an Endpoints field on token records, and ignores /docs/plans/. Callers of the removed internal helpers and components must use the shared equivalents.
Adds passive metadata decoders for AWS, Git, and SFTP wire traffic, which keep signed requests and file or pack contents unmodified. Also adds per-endpoint proxy authorization that checks token scope, endpoint allow-lists, expiry, and revocation on every request, and accepts Basic or Bearer credentials.
Add generated documentation for every trace kind. A KindReference component renders each kind's flags, parameter schema, record and result columns, and example output from docs/_data/kinds.json, which is generated from the Go catalog and checked during the docs build. Adds per-kind MDX pages, a Rspress site configuration, getting-started and replay guides, and a Vitest suite for the component. Links to examples and the README now point at GitHub so they resolve outside the source tree. Process capture documentation describes the unified process, process_facets, and process_metrics tables.
…ic storage

Add a process capture pipeline under pkg/tracing/processcapture that records process identities once at startup, numeric metrics per tick, and resource facet deltas (files, receiving sockets, inbound/outbound connections) in atomic batches. Facet state advances only after storage commits, so failed writes are retried without losing or duplicating deltas. Group totals are derived on read rather than stored as snapshots, and secrets in commands and environment are masked before persistence.
…summaries to tracing

Add a tracing layer that shows what gavel publishes about its tasks: gavel processes are followed by pid through their status files, their task groups and running tasks are nested under them, and their slot pools are summarized. Gavel rules can be skip-listed so their tasks claim processes by pid instead of ancestry. Top also gains compressed and swap memory columns, resource hog detection with a 120s retention, and a grouping by slot pool. Result tables are registered per capture kind, and summaries now report durations, latest prometheus target health, and top-ranked profile functions. Pretty renderers are added for the record kinds.
… and compact run IDs

Merge process_tree and process_samples into a single process kind backed by processcapture, and remove process_groups. Process traces now store identities, metrics and facets in separate streams, and follow reads sequences through the backend instead of the notification stream. Run IDs become a short hostname plus a compact UTC timestamp, and prune orders runs by manifest start time.

BREAKING CHANGE: the process_tree, process_samples and process_groups kinds are removed in favour of process. Run IDs use the form <host><YYYYMMDDTHHMMSSZ>, and collision suffixes use _N instead of -N. Recordings made with the old kinds need the build that produced them to be read.
Add hack/demo, a runnable program that records one persisted Telltale run with real fresh captures for every registered trace kind. It drives localhost HTTP and SQLite orders, a Clicky exec and PTY terminal session of git --version, a child JVM under JFR, managed Prometheus scraping, lab SQL Server XEvents, plan-cache and deadlock captures, and one bounded Captain API LLM request capped at USD 0.05.

Verification fails the run unless every kind has meaningful records with no dropped or unencodable loss, and every session ends stopped or completed. Fixtures are never loaded. The demo also exports Clicky UI playground events and saves the lab connection with a private file mode, refusing to overwrite an existing one.

Also adds hack/docs-catalog, which regenerates docs/_data/kinds.json from the trace registry and can check it for staleness.
Process sampling is now the single kind process, replacing process_samples and process_groups. Host processes reuse the Resources options, and the process capture is renamed to processcapture. Adds a unified per-run records feed (SSE or NDJSON with resumable Last-Event-ID) and result-scoped record selection, plus a schema list helper used by storeSchemas. BREAKING CHANGE: the process_samples and process_groups kinds are removed; definitions must use kind process, and collect.host.processeGroups is replaced by collect.host.processes.
Arm a clicky process's agent probes (exec, entity, task, HTTP, logs, heap) through
one built-in definition so a single selector records its whole workload.
The new entity and task kinds exist only over an agent; telltale registers no such probe itself.
Preset collect groups hold kinds other groups also arm, so an ad-hoc capture lands
there only when no other list takes it.
The records feed gains excludeKind and an optional ANSI presentation, and a process
endpoint returns each identity's latest measurements instead of the paged history.
An empty capture selection now errors only when a kind or stream was requested.
The landing page now lists OpenTelemetry traces and recorded runs together, newest first, with the type, text, source and window filters kept in the URL so a filtered view can be linked and restored.

Opening a run from the overview keeps the shell chrome: navigation, header and action areas are portalled into AppShell slots so a run renders full width without duplicating page headings.

Run detail is split into events, tasks and process details tabs; clicky task captures render through the clicky-ui task viewer, and process facets, stack collection and Go profiles stay available.

The events feed drops process records and renders ANSI, so process data is no longer duplicated outside its own tab.

BREAKING CHANGE: the root page no longer shows the backend connection panel, the embedded trace run list or the OpenAPI link; that content is replaced by the new activity overview.
…al on shutdown

Ship ready-to-run definitions for tracing Java, Go and Node applications and
builds, so users can start profiling without authoring YAML themselves.
The host capture schema now models a single process tree, letting definitions
restrict environment variables and sample resources and network usage per
process; examples and the checkout demo were updated to the new fields.
The CLI now saves and restores terminal state and routes SIGINT/SIGTERM
through a graceful shutdown window, so the alternate screen and cursor are
recovered instead of leaving the shell corrupted after an interrupted run.
Tests cover the process measurement feed, run replay and platform memory
columns to keep the exposed API honest.

BREAKING CHANGE: host.processGroups is replaced by host.processes, the
resources capture is now process-oriented, and the per-query span and
operation field mappings were removed; existing TraceDefinitions must be
migrated.
…o TUI

Signal handling now lives in a single controller that restores the terminal, allows a 30s grace period, and forces exit on a second interrupt, so TUI views and the browser viewer shut down cleanly whether the quit comes from a key or an OS signal.

The trace UI gains an "All events" tab plus a live process header (e toggles exited rows, alt+arrows scroll) fed by a new /traces/{run}/processes endpoint, and the OpenAPI spec and viewer proxy expose that endpoint along with the seq and export operations.

BREAKING CHANGE: TUI sorting now defaults to timestamp instead of seq, and the process kind tab is replaced by the process header.
…unning requests

New sqltop handler polls pg_stat_activity or SQL Server DMVs and emits one row per active request, ranked by elapsed time, with database/text/elapsed filters. Shared connection leasing moves into pkg/tracing/sqlconn so SQL XEvents and sql_top refuse wrong connection types consistently and name the required engine in the error.
…nd SQL Server queries

Poll pg_stat_activity or sys.dm_exec_requests on a connection and record each running request, longest first, with session, wait, blocker, elapsed time and SQL. Sessions holding an open transaction are always recorded; --idle adds idle sessions and --top/--min-elapsed bound the output.

Wire the kind into TraceDefinitions as collect.postgres.top and collect.sqlserver.top, plus the demo coverage predicate, docs page and nav entries.
…so TLS-configured clients keep capturing

Claude-Session-Id: 754971f0-98ba-4fb1-bae2-399104179627
…dates

Bumps the go_modules group with 3 updates in the / directory: [github.com/go-jose/go-jose/v4](https://github.com/go-jose/go-jose), [github.com/moby/go-archive](https://github.com/moby/go-archive) and [github.com/moby/sys/user](https://github.com/moby/sys).
Bumps the go_modules group with 5 updates in the /sdk/captain directory:

| Package | From | To |
| --- | --- | --- |
| [go.opentelemetry.io/otel/sdk](https://github.com/open-telemetry/opentelemetry-go) | `1.44.0` | `1.45.0` |
| [google.golang.org/grpc](https://github.com/grpc/grpc-go) | `1.82.1` | `1.83.2` |
| [github.com/go-jose/go-jose/v4](https://github.com/go-jose/go-jose) | `4.1.4` | `4.1.6` |
| [github.com/labstack/echo/v4](https://github.com/labstack/echo) | `4.15.2` | `4.16.3` |
| [github.com/cert-manager/cert-manager](https://github.com/cert-manager/cert-manager) | `1.20.0` | `1.20.3` |
| [github.com/go-jose/go-jose/v4](https://github.com/go-jose/go-jose) | `4.1.4` | `4.1.6` |
| [github.com/labstack/echo/v4](https://github.com/labstack/echo) | `4.15.2` | `4.16.3` |
| [go.opentelemetry.io/otel/sdk](https://github.com/open-telemetry/opentelemetry-go) | `1.44.0` | `1.45.0` |
| [google.golang.org/grpc](https://github.com/grpc/grpc-go) | `1.82.1` | `1.83.2` |
| [github.com/go-jose/go-jose/v4](https://github.com/go-jose/go-jose) | `4.1.4` | `4.1.6` |
| [github.com/labstack/echo/v4](https://github.com/labstack/echo) | `4.15.2` | `4.16.3` |
| [go.opentelemetry.io/otel/sdk](https://github.com/open-telemetry/opentelemetry-go) | `1.44.0` | `1.45.0` |
| [google.golang.org/grpc](https://github.com/grpc/grpc-go) | `1.82.1` | `1.83.2` |
| [go.opentelemetry.io/otel/sdk](https://github.com/open-telemetry/opentelemetry-go) | `1.44.0` | `1.45.0` |

Bumps the go_modules group with 1 update in the /sdk/commonsdb directory: [github.com/go-jose/go-jose/v4](https://github.com/go-jose/go-jose).
Bumps the go_modules group with 1 update in the /sdk/otel directory: [go.opentelemetry.io/otel/sdk](https://github.com/open-telemetry/opentelemetry-go).


Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/moby/go-archive` from 0.1.0 to 0.3.0
- [Release notes](https://github.com/moby/go-archive/releases)
- [Changelog](https://github.com/moby/go-archive/blob/main/changes_test.go)
- [Commits](moby/go-archive@v0.1.0...v0.3.0)

Updates `github.com/moby/sys/user` from 0.4.0 to 0.4.1
- [Release notes](https://github.com/moby/sys/releases)
- [Commits](moby/sys@user/v0.4.0...user/v0.4.1)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `google.golang.org/grpc` from 1.82.1 to 1.83.2
- [Release notes](https://github.com/grpc/grpc-go/releases)
- [Commits](grpc/grpc-go@v1.82.1...v1.83.2)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/labstack/echo/v4` from 4.15.2 to 4.16.3
- [Release notes](https://github.com/labstack/echo/releases)
- [Changelog](https://github.com/labstack/echo/blob/v4.16.3/CHANGELOG.md)
- [Commits](labstack/echo@v4.15.2...v4.16.3)

Updates `github.com/cert-manager/cert-manager` from 1.20.0 to 1.20.3
- [Release notes](https://github.com/cert-manager/cert-manager/releases)
- [Changelog](https://github.com/cert-manager/cert-manager/blob/master/RELEASE.md)
- [Commits](cert-manager/cert-manager@v1.20.0...v1.20.3)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/labstack/echo/v4` from 4.15.2 to 4.16.3
- [Release notes](https://github.com/labstack/echo/releases)
- [Changelog](https://github.com/labstack/echo/blob/v4.16.3/CHANGELOG.md)
- [Commits](labstack/echo@v4.15.2...v4.16.3)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `google.golang.org/grpc` from 1.82.1 to 1.83.2
- [Release notes](https://github.com/grpc/grpc-go/releases)
- [Commits](grpc/grpc-go@v1.82.1...v1.83.2)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/labstack/echo/v4` from 4.15.2 to 4.16.3
- [Release notes](https://github.com/labstack/echo/releases)
- [Changelog](https://github.com/labstack/echo/blob/v4.16.3/CHANGELOG.md)
- [Commits](labstack/echo@v4.15.2...v4.16.3)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `google.golang.org/grpc` from 1.82.1 to 1.83.2
- [Release notes](https://github.com/grpc/grpc-go/releases)
- [Commits](grpc/grpc-go@v1.82.1...v1.83.2)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `github.com/go-jose/go-jose/v4` from 4.1.4 to 4.1.6
- [Release notes](https://github.com/go-jose/go-jose/releases)
- [Commits](go-jose/go-jose@v4.1.4...v4.1.6)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

Updates `go.opentelemetry.io/otel/sdk` from 1.44.0 to 1.45.0
- [Release notes](https://github.com/open-telemetry/opentelemetry-go/releases)
- [Changelog](https://github.com/open-telemetry/opentelemetry-go/blob/main/CHANGELOG.md)
- [Commits](open-telemetry/opentelemetry-go@v1.44.0...v1.45.0)

---
updated-dependencies:
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/moby/go-archive
  dependency-version: 0.3.0
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/moby/sys/user
  dependency-version: 0.4.1
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: google.golang.org/grpc
  dependency-version: 1.83.2
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/labstack/echo/v4
  dependency-version: 4.16.3
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/cert-manager/cert-manager
  dependency-version: 1.20.3
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/labstack/echo/v4
  dependency-version: 4.16.3
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: google.golang.org/grpc
  dependency-version: 1.83.2
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/labstack/echo/v4
  dependency-version: 4.16.3
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: google.golang.org/grpc
  dependency-version: 1.83.2
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: github.com/go-jose/go-jose/v4
  dependency-version: 4.1.6
  dependency-type: indirect
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: direct:production
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: direct:production
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: direct:production
  dependency-group: go_modules
- dependency-name: go.opentelemetry.io/otel/sdk
  dependency-version: 1.45.0
  dependency-type: direct:production
  dependency-group: go_modules
...

Signed-off-by: dependabot[bot] <support@github.com>
…runs

Claude-Session-Id: f3377f49-2d4e-4a63-afd9-a71b38cde52e
Bumps the go_modules group with 1 update in the / directory: [github.com/xuri/excelize/v2](https://github.com/xuri/excelize).


Updates `github.com/xuri/excelize/v2` from 2.11.1-0.20261029001137-c5fd73bde8f6 to 2.11.1
- [Release notes](https://github.com/xuri/excelize/releases)
- [Commits](https://github.com/xuri/excelize/commits/v2.11.1)

---
updated-dependencies:
- dependency-name: github.com/xuri/excelize/v2
  dependency-version: 2.11.1
  dependency-type: indirect
  dependency-group: go_modules
...

Signed-off-by: dependabot[bot] <support@github.com>
Expose sampled trace captures through a Top view so clients can read the latest complete sample, with native filters, sort order, paging, and profile metrics applied before pagination. Completed sample frames can also be listed. The records endpoint now accepts frame, source, metric, sort, and desc parameters, and the OpenAPI spec documents the new records path and query options. Requests for samples without complete boundaries return 409 instead of stale rows, and the SQL and profile kinds advertise Top support with connection lookup for the web start form.
Tracing kinds (goprof, jfr, stacks, prometheus, plancache, sqltop, processcapture) now publish sample batches as atomic frames via tracing.PublishFrame, so each frame's rows and its summary (observed count, limit, metrics) are written together. Kinds that declare a TopSpec get a Top view built from those frames, and the kind schema is validated against it at registration. The sqltop readers and the postgres/sqlserver query code move into commons-db query/sqlactivity, and sqltop.Sample becomes an alias of it. BREAKING CHANGE: sqltop no longer exports its readers or commandOf helper, and emitters used by sampled kinds must support EmitFrame. The SDK gains Unwrap/Rewrap on the HTTP transport so commons clients that replace http.DefaultTransport keep capture, and the captain SDK moves to the observe package.
Adds a Top tab for sampled captures with paging, filters, search, and linked row lookups (e.g. blockers), pinned to a completed sample and restored from URL state. The Events tab is renamed History and Top becomes the default run view. BREAKING CHANGE: runs now open on Top by default and the Events tab is renamed History. Also allows clicky-ui 0.3.58 in minimumReleaseAgeExclude and links clicky-ui to the local checkout via pnpm overrides.
…ure kinds

Adds JVM agent-seam captures for JMX notifications and GC, periodic OGNL sampling, method watch/trace, and class redefinition with restore on capture end. Each kind records its target, and the agent seam now treats a Complete end as success instead of an error. Source paging is fixed to sample a stable write cursor so busy producers no longer skip or stall pages. Definitions, CRD schema, and an example fixture are updated to match.
… hub

Expose the live Java agent catalog, generic operation calls, one-shot Groovy evaluation, and diagnostic extension list/install/replace/unload through the CLI and the /api/agents/{agent} HTTP endpoints. Mutating calls require --allow-mutation and are refused unless the agent reports a Java runtime. Flags can now accept JSON arrays of objects for kind params, and /jvm/ paths serve the UI. Adds the agent, traces and jvm routes to the OpenAPI spec.
…nd instrumentation

Add a JVM diagnostics page to the webapp covering JMX, classes and source, OGNL and Groovy evaluation, threads, metrics, logging, bytecode patches, method instrumentation, traces and JMX notifications. Method captures and notification captures reuse the trace run pipeline and record the target agent so recordings are scoped to a single JVM. Mutating operations require explicit per-action consent, and redefinitions are leased so original bytecode is restored automatically. The agent API gains a catalog endpoint, operation calls with allowMutation, and extension install and unload, and capture targets now carry allowMutation.
… arm page

Let users attach the Java agent to a JVM on the server host or deploy it into Kubernetes pods from the agents page, the arm wizard, and capture pickers. The arm page is rebuilt as a step wizard where each capture picks its own target (connection, agents, proxy, or process), and a kind can be armed more than once. Start requests send a list of entries for kinds armed multiple times, and the trace entry type reflects that. Captures previously configured in a single form now go through the new wizard model with server-validated refusals mapped back to their step.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
…er connection type

Trace kinds can now take an array of capture objects, each with its own target and params, and refusals are reported per element index. The SQL Server deadlock and xevent handlers now declare the SQL Server connection type so they are matched only to compatible connections. The docs describe the array form.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
…untime mutation gating

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
…m the profile read route

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Adds a support snapshot that bundles JVM diagnostics (settings, resources, flags, metrics, thread dump, class loaders, logging, status, optional heap histogram and a JFR recording) into a zip with a manifest and summary. Each artifact is scrubbed for secrets in properties, arguments, and URLs, and collectors are bounded by a 30s timeout and 64 MiB/128 MiB size limits. A failing collector is recorded in the manifest instead of aborting the snapshot. JFR scrubbing requires the JDK 19+ jfr tool on the Telltale host, and the target JVM can be resolved by connected agent or local PID.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Adds a redacted JVM support ZIP via the CLI (local PID or remote hub) and a POST /api/v1/jvm/snapshot endpoint that streams the bundle and reports completion status in a header. Partial bundles are flagged so operators can see collection failures.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
…nostics

Expose read-only JVM settings, Linux resource and cgroup limits, effective HotSpot flags, and a heap histogram through the diagnostics API so operators can inspect runtime configuration without mutation. The heap histogram includes unreachable objects and may pause the JVM. Also adds a support snapshot option to JFR recordings that limits enabled events and caps max size at 60 MiB.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Lets operators collect a support bundle from a JVM agent from the JVM page. Expensive collectors (JFR profiling and heap histogram) stay opt-in, partial bundles are flagged, and in-flight collection can be cancelled.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Adds telltale jvm snapshot to bundle diagnostic evidence into a ZIP and exposes typed commands for built-in JVM operations, replacing generic jvm call examples in the docs. Adds jvm.settings, jvm.resources, jvm.flags and jvm.heap.histogram to the catalog snapshot, which must be regenerated with task generate when the Java registry changes. Existing operations and jvm call remain available for extension operations.

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
… of the operation count

Claude-Session-Id: cce50d69-7670-4bcb-b837-c8cbf8519c2f
Add a JVM Top command and web tab that sample CPU, memory, GC and threads from one connected Java runtime, and a native activity telemetry read that reports loaded applications without arming them or using JMX. The support snapshot gains workload.html, sampled over a configurable window, and telemetry.html, and the snapshot timeout rises to 12m to cover sampling. Snapshot redaction now masks camelCase secret keys such as trustStorePassword, and agent resolution for snapshots goes through Lookup so deployed targets are found. Operators get live diagnostics in the CLI, API and web UI, with read-only collection of new data.

Claude-Session-Id: 01a1e7f9-495a-7f02-91dd-1575da68bb2d
Saved deployer targets are now listed by the agents CLI, HTTP API, and web UI alongside connected agents, grouped by namespace, cloud, or host. Inventory reads saved metadata only and does not connect targets. The API now returns agentclient.Info instead of protocol.Hello, so clients must read agentId, connected, group, and known fields. The shared terminal rendering helpers moved from the top UI into terminalui.

Claude-Session-Id: 01a1e7f9-495a-7f02-91dd-1575da68bb2d
Document and wire up native activity telemetry in JVM snapshots, a workload panel sampled every 5s, and the new jvm top view. The snapshot API gains an optional workload field, and signing-connection now reuses deploy.WriteIssuer so the key logic is shared with jvm deploy.
Comment thread webapp/vite.config.ts Fixed
Comment thread hack/jvm-demo/java/DemoServer.java Fixed
Comment thread internal/jvm/snapshot/collect.go Fixed
Local Clicky UI aliases only replaced the first wildcard and checked the wrong directory, so subpaths like themes/* failed to resolve. The asset check now uses the directory before the wildcard, and every wildcard in the target is substituted.

Gavel-Issue-Id: 4e3fad86-67af-47d3-83c9-475bd359cba2
Claude-Session-Id: 01a1efe4-84c8-7251-9dd0-431c6d9fbb2f
… scrape captures

Replace the managed Prometheus with a managed VictoriaMetrics instance. Captures now report target health and store samples directly in VictoriaMetrics for 15 days, instead of emitting one scrape record per completed scrape. Add serve --prometheus to keep the instance running between runs in the same data directory. Previous Prometheus state under services/prometheus is left in place and not migrated, so earlier samples are not queryable from the new instance. BREAKING CHANGE: scrape records are no longer emitted, the data directory layout moves to services/victoriametrics and victoriametrics/targets, and the deps/start dependency is removed.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants