Observability plugins

Instrument Go LLM applications with the neatlogs SDK — Gemini via WrapGenAI, Google ADK via the private-provider integration, explicit span helpers, custom boundaries, and per-request identity.

23 skills

Connect Claude Code to Cardinal: telemetry to the Outcomes Dashboard. Unified Cardinal MCP coming in a future release.

11 skills · 2 agents · 2 MCP servers · hooks

Query and analyze SparkLogs: system and application logs plus device health and state over time, one host or the fleet.

4 skills · 2 commands · 3 agents · 1 MCP server

Observability concept plugin: RED/USE + logging + tracing design AND Datadog/Grafana/Honeycomb/Elastic/Sentry query integration.

9 skills

Baseline karmaIQ MCP connection, active-domain picker, exploring-service-mesh skill, and help/discoverability skill. Install first — required by all other karmaIQ plugins.

5 skills · 1 MCP server · hooks

Structural risk review across the karmaIQ service mesh — cycles, single points of failure, orphan services, dead code, hot methods, fan analysis. Requires karmaiq-core.

6 skills · 1 agent

Preview workflows for Sentry: open fix PRs and groom stale issues. Built on the official Sentry MCP.

3 skills · 2 agents · 1 MCP server

Flowlines MCP observability: analyse MCP server usage, investigate sessions, and instrument MCP servers.

5 skills · 1 MCP server

Autonomous SRE for production incident root-cause analysis via karmaIQ. Pre-warms service lookups when you mention APIs or services. Requires karmaiq-core.

4 skills · 1 agent · hooks

Read your production logs from inside your coding agent. MCP server plus the investigation and SDK integration skills.

2 skills · 2 commands · 1 MCP server

Investigate production incidents from live Grafana data (metrics, logs, traces) with a trust-check-first procedure, and keep alerts and dashboards as code with a linter, publisher and live verifier.

3 skills · 1 agent

Captures Claude Code session events (sessions, prompts, tool usage) into a local SQLite store so future versions can detect patterns and forge skills/agents/hooks automatically. PR1 is capture-only; ships the /insights skill as a smoke-test reader.

2 skills · hooks

Full-spectrum agent observability: hooks every Claude Code lifecycle event and emits structured JSONL events

1 skill · 1 MCP server · hooks

Search BlazeMeter logs across denv/bzdev/ci/staging/prod — bundles a read-only opensearch-reader MCP server plus a skill that builds safe Lucene queries and condenses results.

1 skill · 1 MCP server

Claude Code plugin for the Switch Trust platform. Reports Claude Code usage & computed cost to Switch Trust for cost tracking and further analysis.

1 command · hooks

Push notifications for Claude Code: ntfy.sh, macOS native, and Pushover with sound themes. Frozen at the archived Agentic Primitives copy until it gets an active home.

1 command · hooks

Respan for Claude Code: the Respan skill and the hosted Respan MCP server, for traces, evals, prompts, datasets, and gateway routing.

1 skill · 1 MCP server

Logs token usage for every subagent (Task tool) call to ~/.claude/subagent-metrics.jsonl

hooks

Integration playbook for the @darkhunt-security/telemetry SDK — wires LLM tracing, generations, and spans into a Node/TypeScript service. Auto-invokes when you ask Claude to add Darkhunt telemetry / observability to a service.

1 skill

Integration playbook for the darkhunt-telemetry Python SDK — wires LLM tracing, generations, and spans into a Python service. Auto-invokes when you ask Claude to add Darkhunt telemetry / observability to a Python service.

1 skill

Structured playbook for investigating a misbehaving BlazeMeter master/test/session (stuck, wrong metric, silent failure, intermittent, customer-reported) across denv/bzdev/ci/staging/prod — composes OpenSearch log search, Mongo (main + bigdata) state, GCS session artifacts, and source-code verification across taurus/taurus-cloud/a.blazemeter.com/bzm-crane.

1 skill

Investigate agent runs recorded in Langfuse: failed sessions, token usage, tool-call patterns.

1 skill

Track per-device Claude Code usage, rejection reasons, and per-session stats from a single dashboard.

hooks

OpenObserve REST API skill for AI agents. Search logs/metrics/traces, manage streams, and create/edit/inspect dashboards via curl.

1 skill

Read-only cross-repository API impact evidence for the current checkout: workspace status, change analysis, consumer lookup, API flows, function usage, and fix verification over MCP.

A live agent dashboard pane: model vitals, every permission check, subagent cards and a session log. Observes only. Reviewed: Recommend.

A small box above the prompt that always says what Claude is doing right now, and expands to show what it has done. Reviewed: Recommend.

One live line above the prompt per running subagent: time, tool count and what it's doing. /radar shows every agent and its messages. Reviewed: Recommend.

Send your Claude Code sessions to Respan as traces, with a span for each model request and tool call.