dev.agentreliability/agent-reliability
Agent Reliability
Testing, benchmarking and auditing autonomous AI agents — methods, harnesses, evidence
alive probed 2026-10-14 · HTTP 200
9tools
3.2ktokens in Claude Opus 5.5
3.3ktokens in Claude Code
1.6 %of a 200k context window
2.0ko200k_base estimate
1.6 sto initialize
257 msto list tools
- Endpoint
- https://agentreliability.dev/mcp?via=manifest (streamable-http)
- Host
- agentreliability.dev
- Protocol
- 2025-11-25
- Reports itself as
- dev.agentreliability/agent-reliability 0.9.2
- Capabilities
- completions, resources, tools
- Instructions
- none
- Claude count
- Claude Opus 5.5, measured 2026-10-05. Counted again when the tool definitions change. Method.
- Repository
- https://github.com/citarium/agentreliability-mcp
- Registry
- active, updated 2026-08-23 · registry entry
- Reachable
- 7 of the last 7 probes
Most expensive tools
| Tool | Tokens (o200k) | Share of this server |
|---|---|---|
| get_sources | 389 | 20 % |
| search | 315 | 16 % |
| get_topic | 281 | 14 % |
| get_related | 251 | 13 % |
| compare | 228 | 12 % |
History
We add a row when the result changes. A server with no change keeps its last row.
| Probed | Result | Init | Tools | Claude | o200k |
|---|---|---|---|---|---|
| 2026-10-14 18:26 | alive | 2.8 s | 9 | 3.2k | 2.0k |
| 2026-10-05 13:56 | alive | 1.6 s | 9 | 3.2k | 2.0k |
| 2026-10-05 00:24 | alive | 960 ms | 9 | 2.0k | |
| 2026-10-04 10:00 | alive | 1.7 s | 9 | 2.0k | |
| 2026-10-04 07:53 | alive | 2.4 s | 9 | 2.0k | |
| 2026-10-03 20:33 | alive | 2.3 s | 9 | 2.0k | |
| 2026-10-03 17:50 | alive | 1.1 s | 9 | 2.0k |
Badge and feed
[](https://mcpcost.ulehla.dev/s/dev.agentreliability/agent-reliability)
An entry for every probe that changed the result, the tool count or the token cost: Atom feed. Add it to a feed reader or a Slack RSS app to hear when the server stops answering or its tool set grows.