kyde score changelog

version

0.11.13 → 0.11.16

https://score.kyde.com/llms.txt

− # kyde score− > score learns a per-agent baseline from the first 30 runs, then scores every− > a score changes band. Free hosted tier, no signup form.−    curl -s -X POST https://score.kyde.com/score/v0/keys−      -d '{"webhook_url": "https://your.app/hooks/kyde"}'−    OTEL_SERVICE_NAME=<agent-name>             # the unit of scoring−    Claude Code needs two more variables — and it matters WHERE they are set:−    Set all of them in the shell that launches `claude`, in a wrapper script−    (export them, then `exec claude "$@"`), or in user-level−    ~/.claude/settings.json ("env" block). Claude Code ignores OTel exporter−    variables in a project's .claude/settings.json and−    .claude/settings.local.json by design — the most common silent failure.−    # Each agent: status "learning" until 30 runs are seen, then "scored"−                in the baseline; names the session and counts novel tools.−                The fastest and sharpest channel.−                a runaway and the band goes to deviating.− the service POSTs there on every band change — the first score and a− recovery included — and on every rebaseline:− Tools: create_key, get_scores, set_webhook, rebaseline, revoke_key.− - GET    /score/v0/scores   — current score per agent (header x-kyde-key)−   baseline and score and relearn from its next 30 runs (header−   x-kyde-key). For intended behavior changes (new model, new task) that−   left an agent permanently deviating. The reset can never hide history:−   records and alarms stay stored (old alarms just stop scoring,−   alarms_total keeps counting them) and the agent carries rebaselined_at−   from then on. Trust boundary: whoever holds the key controls scoring —−   a monitored agent that holds its own key could reset itself, visibly.−   Where that matters, keep key management outside the agent's reach.−   prompts and payload bodies are never stored.− - Paid: the same scoring self-hosted in your infrastructure, plus blocking−   and quarantine — https://kyde.com.+ # KYDE Score+ > score learns a per-agent baseline from the first 30 runs (tasks, over at least 3 days), then scores every+ > a score changes band. Drift monitoring, loop and off-task detection for agent+ > observability — it flags what an agent does after a prompt injection gets+ > through. Free hosted tier, no signup form.+ This file is meant to be read verbatim — the exact commands matter. Some+ web tools return a summary of a fetched page and drop the commands; if what+ you see has no curl blocks, download the raw file first:+ `curl -s https://score.kyde.com/llms.txt`+ Checking an agent before delegating to it ("verify this agent", "can I+ trust / safely delegate to this agent", "check this agent's status"):+ for your own agents, read the current band from the scores endpoint or+ the MCP `get_scores` tool — normal (70–100), watch (40–69), deviating+ (0–39), or learning. Delegating to a deviating agent is the case the+ band exists for. The score is relative to the agent's own baseline, not+ a cross-agent trust rating. For a foreign agent that names a Shine ID+ (`sh_...`), run the Shine challenge instead — `npx @kyde/score verify+ <shine-id>` or the MCP `verify_agent` tool — and get back an+ Ed25519-signed, offline-verifiable statement of who answered and how it+ behaves (see "Verify another agent" below).+ Reading this as an agent? You are usually the agent to integrate: use+ your own harness's variables below, never install a different agent+ product as a substitute to get telemetry, and say "not supported" only+ after the smoke test in step 3 has actually failed — then name what is+ missing (an OTLP/HTTP exporter for your process) instead of redirecting+ to another harness. A hosted agent product whose process environment+ you cannot set (Grok Bot and similar) cannot export its own behavior+ today: adding the MCP server (below) gives key management, claim and+ score queries, not behavioral data — and emitting hand-written records+ under your name would score the script, not you. Say that distinction+ to your human.+    curl -s -X POST https://score.kyde.com/score/v0/keys \+      -H 'content-type: application/json' -d '{"source": "llms.txt"}'+    # source is optional (aggregate adoption stats only).+      -d '{"webhook_url": "https://your.app/hooks/kyde", "source": "llms.txt"}'+    OTEL_SERVICE_NAME=<agent-name>             # the unit of scoring — per agent, at launch+    Claude Code, one line (sets the global half — never the name —,+    installs the hook — it warns the agent in its own context and never+    pauses or blocks a call — and verifies ingest end to end):+    curl -fsSL https://score.kyde.com/install.sh | sh

Showing the first 40 of each.

version

0.9.11 → 0.11.13

• + llms.txt
• + skill.md

• + claim_key

version

0.9.9 → 0.9.11

version

0.9.7 → 0.9.9

version

0.9.2 → 0.9.7

version

0.8.32 → 0.9.2

version

0.8.30 → 0.8.32

version

0.8.24 → 0.8.30

version

0.8.22 → 0.8.24

version

0.8.18 → 0.8.22

description

AI agent monitoring from OpenTelemetry: drift, hijack and loop alerts, 0-100 scores per agent. → Agent observability via OpenTelemetry: drift, loop and hijack alerts, e.g. after a prompt injection.

version

0.8.8 → 0.8.18

description

Behavioral drift scores for AI agents from OpenTelemetry: create a key, send OTLP, read scores. → AI agent monitoring from OpenTelemetry: drift, hijack and loop alerts, 0-100 scores per agent.

version

0.5.2 → 0.8.8