CompletionKit
com.completionkit/evals
Prompt evals over MCP: run a prompt on your dataset, score each output 1-5 with an LLM judge.
— the operator's own registry description reported
Publisher com.completionkit · first seen 2026-08-12 · endpoint https://completionkit.com/mcp
Dependency rating derived
Cluster stale-rich-solo · distinctiveness 0.397 (moderate). Every figure is a percentile or class within the 11,094-server census — see the whole spectrum.
🔏 Signed receipt rr_bea1cf6df5cb38baf4e2f236 · ed25519 · key rs-rcpt-2026-08 — this rating is tamper-evident; an agent gets the full signature from check_server and can verify it against the published key.
What kind of tools these are derived
Verification only means something relative to a tool's NATURE — an email-sender has no "true answer", a generator has no answer key. This server's 54 tools, classified, with what checking applies to each kind (the full methodology →):
| nature | tools | meaning | what checking applies |
|---|---|---|---|
| action | 39 | changes the world | no true answer exists; we check the CONTRACT (destructive/read-only/idempotent declarations, error legibility) and never fire real actions |
| retrieval-public | 14 | serves public facts | truth-checkable against a public source — the V-ladder applies in full |
| generative | 1 | creates content | no answer key exists; disclosure + stability checks, never truth verdicts |
Tools last seen
Harness readiness observed
Can an agent's harness pick this tool and call it correctly? Graded on the three things an agent reads — name, description, typed parameters. Band B — usable, with rough edges.
| legibility check | tools passing | ok |
|---|---|---|
| name is clear & specific | 54/54 | ✓ |
| has a description | 54/54 | ✓ |
| description is substantive | 27/54 | 27/54 |
| parameters are typed | 54/54 | ✓ |
| parameters are described | 11/54 | 11/54 |
54 of 54 tools graded · 54 with a captured input schema. RS-008, Tier-1 — no domain knowledge, applies to any tool. “Alive” is not the same as “usable”.
Truth checks observed
Not in the Tier-2 trade lane, or not yet checked. Tier-2 re-derives a server's answers against published primary sources — see truth checks.
Protocol conformance observed
Capabilities: tools
Drift
No confirmed drift on record. Baseline set 2026-08-14; changes appear here after human review.
History observed
Every probe we made, oldest → newest (2 shown, 2026-08-12 → 2026-08-14). Green answered · amber answered-but-walled · red no useful answer. A gap in our cadence is a gap in coverage, not evidence about the server.
changelog (1 events)
| date | type | what changed |
|---|---|---|
| 2026-08-12 | outcome | first probe: ok_tools |
raw probe records
| probed (UTC) | outcome | http | ms | protocol |
|---|---|---|---|---|
| 2026-08-14T19:58:01Z | ok_tools | 200 | 127 | 2025-03-26 |
| 2026-08-12T05:33:17Z | ok_tools | 200 | 145 | 2025-03-26 |
Watch or embed this verdict
Operators: put the live verdict in your README — it updates with every census, links back here, and is a dated observation, never a warranty:
[](https://robinsaige.com/s/com.completionkit/evals)
Depend on it? Subscribe to its change feed — outcome changes and confirmed drift, no account needed. Building on it? The full dossier as JSON — verdict, rating, truth checks, history — stable enough to gate CI on.