Grafana Agent Observability is an SDK and plugin collection that collects telemetry from coding agents and AI agents built into applications, including sessions, tool calls, traces, tokens, costs, generations, and evaluations. Developers use it to monitor coding-agent usage or instrument agents in their own services, with SDKs for several programming languages. The catalogue add-ons help configure and use this observability workflow with coding agents.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add grafana/agento11y --skill e2e-testgit clone --depth 1 https://github.com/grafana/agento11yWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/grafana/agento11y/e2e-test)<a href="https://agentmods.dev/skills/grafana/agento11y/e2e-test"><img src="https://agentmods.dev/badge/skills/grafana/agento11y/e2e-test/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/grafana/agento11y/e2e-test"><img src="https://agentmods.dev/badge/skills/grafana/agento11y/e2e-test.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00042 | $0.01733 |
| Opus 5.5 | $0.00017 | $0.00693 |
| Sonnet 5.5 | $0.00008 | $0.00347 |
| Haiku 4.5 | $0.00004 | $0.00173 |
Grade A, and why
e2e-test scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 22d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Optional Hermes end-to-end checks
Unit tests stub the SDK and invent hook payloads. These optional checks inspect real Hermes hooks and exported OTel data. They are not part of normal CI.
Run from plugins/hermes. Read each script before execution. Do not run Cloud
verification, install packages, start servers, or invoke Hermes without approval.
Safety boundary
A local telemetry sink does not make the model provider local. A mock provider does not make telemetry local either. Explicitly configure all three destinations:
| Destination | Credential-free choice |
|---|---|
| Model provider | LOOPBACK: openai-api, mock-model, http://127.0.0.1:8799/v1, dummy key |
| Generation export | Disabled: AGENTO11Y_PROTOCOL=none, AGENTO11Y_AUTH_MODE=none, no token or tenant |
| Traces and metrics | LOOPBACK: http://127.0.0.1:8801, no auth headers |
Use a new temporary HOME and HERMES_HOME, not the user's real config. Clear the
inherited environment so provider keys, telemetry aliases, per-signal endpoints,
proxy settings, and AGENTO11Y_ENV_FILE cannot redirect the test.
Hermes's .env overrides process exports: never reuse an old test home.
Use synthetic content only. Hook-probe logs contain payloads before redaction.
Wrapper limitations
setup.shinstalls packages and defaults its config to Anthropic. Its warning requesting an Anthropic key is not applicable to the loopback recipe.run-hermes.sh sinkredirects OTel only. Its provider defaults to Anthropic, sosinkalone is neither credential-free nor a local-only test.run-hermes.sh fullleaves capture mode unset. It testsmetadata_only, despite its name. The wrapper does not pass through capture mode, redaction, or automatic-tag environment switches.run-mock.shchooses a loopback provider but reads generation and OTLP destinations from the environment orAGENTO11Y_ENV_FILE. It forces basic auth and uses broadpkillmatching. Do not use it for this recipe.verify-backend.shuses Cloud queries. It is outside this credential-free flow.otlp-sink.pydecodes spans and metrics, but logs only attribute names. It does not validate generation ingestion or secret-redaction values.
What ships with it
12 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- scripts/check-generations.py 4.5 KB runs code
- scripts/check-install.py 1.3 KB runs code
- scripts/mock-provider.py 8.2 KB runs code
- scripts/otlp-sink.py 3.4 KB runs code
- scripts/probe-plugin/__init__.py 2.5 KB runs code
- scripts/probe-plugin/plugin.yaml 351 B
- scripts/run-hermes.sh 5.1 KB runs code
- scripts/run-mock.sh 3.1 KB runs code
- scripts/setup.sh 2.0 KB runs code
- scripts/show-hooks.py 2.3 KB runs code
- scripts/show-spans.py 2.2 KB runs code
- scripts/verify-backend.sh 3.2 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 22d ago First seen · 127 lines · 42 tokens per session scan A 6510dabc6dec
e2e-test is a skill published in the GitHub repository grafana/agento11y (129 stars, last pushed today), licensed Apache-2.0. It adds 42 tokens to every session and 1,733 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-24.
Other skills, from other repositories
eval-dataset-design
A guide to creating reliable test sets for AI agents. Each evaluation task includes the request, the simulated user's behaviour, a reset starting state, and a success check that can be independently verified.
test-master
Generates test files, creates mocking strategies, analyzes code coverage, designs test architectures, and produces test plans and defect reports across functional, performance, and security testing disciplines. Use when writing unit tests, integration tests, or E2E tests; creating test strategies or automation…
verify-omnigent
Drive Omnigent the way a user does and prove a behavior with recorded evidence, using an isolated server, runner, and mock model plus a feature map of every user entry point. Load before reproducing a user-facing bug, before claiming a fix works, or when reviewing whether a change covered every surface (session vs.…
dogfood
Systematically explore and test a mobile app on iOS/Android with agent-device to find bugs, UX issues, and other problems. Use when asked to dogfood, QA, exploratory test, find issues, bug hunt, or test this app on mobile.
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
critique
Adversarial design critique of the current work — render it, look at it, and argue for rejection. Run after the gates are green, never instead of them.