Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user asks Codex to inspect Aient telemetry, search logs, explain traces, create/list Aient API keys, or use the Aient MCP server.
.claude/skills/hashgraph-online-aient/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 735% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 105% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 650% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -29% | 0% |
Use this skill when a user asks for Aient operational context from Codex. The plugin exposes the production Aient MCP server at https://aient.ai/mcp. The MCP server is the source of truth for tool schemas; discover/use the live tools rather than relying on a memorized subset.
Start with verify_connection when checking whether the plugin is available. It confirms the organisation, granted scopes, token expiry, and live tool count.
Prefer the current granular tools over the retired early-preview names telemetry_overview, log_search, and trace_timeline.
Connection and discovery:
verify_connection: confirm auth and available MCP tools.describe_telemetry: discover services, operations, and attributes.service_health_summary: get an at-a-glance service health overview.Telemetry:
query_traces: aggregate trace metrics such as count, error rate, and latency.query_logs: aggregate log metrics by service, severity, or time.get_spans: fetch raw spans for a trace or filtered operation set.get_logs: fetch raw log entries and attributes.detect_anomalies: detect statistical deviations in trace/log metrics.Problems and lifecycle:
list_problems: list detected problems with status, severity, priority, andenvironment filters.
get_problem: inspect a problem, recent occurrences, and triage context.acknowledge_problem, mute_problem, unmute_problem,dismiss_problem, resolve_problem, restore_problem_status, update_problem_priority, and batch_problem_action: apply lifecycle actions to one or more problems.
Remediation:
request_fix: spawn an AI remediation agent for a problem.get_problem_fix: inspect a remediation attempt.list_problem_fixes: list remediation attempts for a problem.list_active_fixes: list in-progress remediation work across theorganisation.
Threads:
get_thread: inspect a thread journal and current agent activity.send_thread_message: send a message to a thread's agent.respond_to_interaction: answer an agent interaction/awakeable.Environment and key management:
list_environments: inspect deployment environments.list_environment_keys, create_environment_key, revoke_environment_key:manage publishable telemetry ingest keys.
list_api_keys, create_api_key: manage secret API keys for sourcemaps orcomplete API access. create_api_key returns the full key exactly once.
a local/dev environment.
plans, or plugin files.
aient.mcp.read; mutating tools requireaient.mcp.write.
create_api_key, tell the user that the returned key is shown once andshould be stored immediately.
environmentSlug over environmentId when theuser names an environment, and never pass the all-zero UUID placeholder.
401 as missing or expired OAuth, 403 as insufficient scope, and amissing organisation claim as an auth-context issue.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,461 | 14,376 | -7% | 1 | 1 | 0% | 1,511 | 1,045 | -31% | 0 | 0 | — |
case-02 | fail→fail | 5,645 | 15,170 | +169% | 1 | 1 | 0% | 913 | 1,100 | +20% | 0 | 0 | — |
case-03 | fail→fail | 18,715 | 17,728 | -5% | 1 | 1 | 0% | 1,831 | 1,320 | -28% | 0 | 0 | — |
case-04 | pass→pass | 19,144 | 15,336 | -20% | 1 | 1 | 0% | 2,443 | 2,310 | -5% | 0 | 0 | — |
case-05 | pass→pass | 14,906 | 14,043 | -6% | 1 | 1 | 0% | 1,925 | 2,169 | +13% | 0 | 0 | — |
case-06 | pass→pass | 10,096 | 14,205 | +41% | 1 | 1 | 0% | 1,640 | 2,335 | +42% | 0 | 0 | — |
case-07 | fail→pass | 14,722 | 11,363 | -23% | 1 | 1 | 0% | 233 | 1,945 | +735% | 0 | 0 | — |
case-08 | fail→pass | 6,056 | 13,716 | +126% | 1 | 1 | 0% | 963 | 1,978 | +105% | 0 | 0 | — |
case-09 | fail→pass | 9,276 | 8,192 | -12% | 1 | 1 | 0% | 287 | 2,153 | +650% | 0 | 0 | — |
case-10 | fail→fail | 23,515 | 14,228 | -39% | 1 | 1 | 0% | 1,910 | 1,243 | -35% | 0 | 0 | — |
case-11 | fail→fail | 20,984 | 8,469 | -60% | 1 | 1 | 0% | 2,583 | 1,245 | -52% | 0 | 0 | — |
case-12 | fail→pass | 13,687 | 3,268 | -76% | 1 | 1 | 0% | 2,260 | 1,386 | -39% | 0 | 0 | — |
case-13 | fail→fail | 18,244 | 3,788 | -79% | 1 | 1 | 0% | 1,981 | 1,430 | -28% | 0 | 0 | — |
case-14 | fail→fail | 6,768 | 14,103 | +108% | 1 | 1 | 0% | 368 | 1,453 | +295% | 0 | 0 | — |
case-19 | pass→pass | 12,456 | 7,924 | -36% | 1 | 1 | 0% | 1,225 | 1,302 | +6% | 0 | 0 | — |
case-15 | fail→fail | 9,970 | 25,410 | +155% | 1 | 1 | 0% | 664 | 1,182 | +78% | 0 | 0 | — |
case-16 | fail→pass | 34,746 | 9,823 | -72% | 1 | 1 | 0% | 1,903 | 1,346 | -29% | 0 | 0 | — |
case-17 | pass→pass | 14,203 | 11,154 | -21% | 1 | 1 | 0% | 1,422 | 1,668 | +17% | 0 | 0 | — |
case-18 | pass→pass | 17,082 | 8,508 | -50% | 1 | 1 | 0% | 1,918 | 1,361 | -29% | 0 | 0 | — |
case-20 | fail→pass | 13,859 | 3,194 | -77% | 1 | 1 | 0% | 1,350 | 1,328 | -2% | 0 | 0 | — |
case-21 | fail→pass | 12,874 | 10,932 | -15% | 1 | 1 | 0% | 2,200 | 1,651 | -25% | 0 | 0 | — |
case-22 | fail→fail | 12,764 | 2,736 | -79% | 1 | 1 | 0% | 1,766 | 1,269 | -28% | 0 | 0 | — |
case-23 | fail→pass | 7,477 | 7,109 | -5% | 1 | 1 | 0% | 1,183 | 1,105 | -7% | 0 | 0 | — |
case-24 | fail→pass | 19,182 | 3,268 | -83% | 1 | 1 | 0% | 2,227 | 1,364 | -39% | 0 | 0 | — |
case-25 | fail→pass | 12,654 | 6,045 | -52% | 1 | 1 | 0% | 1,168 | 1,678 | +44% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted, and 17 counted toward the lift figure. The other 8 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +40 percentage points is the difference between those two pass rates over the 17 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.