agentleFS
Sign inStart free
IgorGanapolskyIgorGanapolsky/ThumbGate/.agents/skills/datadog-llm-obs-compare-not-clone/SKILL.md

datadog-llm-obs-compare-not-clone

Skill29 starsChanged 7 months agoMITView on GitHub

Datadog LLM Observability eBook is FORMAT, not a ThumbGate clone. Steal four practices (operational metrics, injection+PII scrub, quality evals, e2e parented spans) onto existing receipts/redaction/feedback/audit-trace. Never vendor dd-trace, DatadogAgentObservability, or OTLP. Never dual-edit PR #3881. Slash: /datadog-llm-obs-compare-not-clone.

  • Reads credentials
  • Installs packages
Create a free account to mount
Files
SKILL.md

Datadog LLM-obs — compare, do not clone

Goal

Produce fail-closed honesty for whom: ThumbGate agents claiming an LLM/agent run is observed — so errors, latency, tokens, injection/PII, quality, and parented spans exist on existing rails before any "we have observability" claim.

Constraints

NEVERALWAYS
Clone Datadog Agent Observability / SDS / ClustersMap 4 practices onto existing rails
npm install dd-trace / DD_API_KEY LLM obsscripts/secret-redaction.js + agent-audit-trace.js
Dual-edit PR #3881 Datadog-style engineComplement; orchestrator overlap is board-loop #3883
Dual-edit PR #3883 board-loopLeave PR drain there
OTLP-export prompt tracesLocal receipts + redaction
Quote Datadog MTTR as oursMeasure ThumbGate latency-budget / task-outcomes
Hero Continuity / net-new obs SKUECI: maintenance of existing surfaces

HARD fail closed. REFUSE SKU clones.

Reference

Examples (show, don't tell)

Weak: Summarize the eBook and add a DatadogAgentObservability class.

Gold:

$ npx thumbgate llm-obs-honesty --json
ok: true
status: ready
$ npx thumbgate llm-obs-honesty --clone-datadog --json
datadog_clone_refused

Procedures

npx thumbgate llm-obs-honesty --json
npx thumbgate llm-obs-honesty --map-only
npx thumbgate llm-obs-honesty --trace=tests/fixtures/llm-obs-honesty-blind.json --json
npm run test:llm-obs-honesty
  1. Require operational counts + latency + tokens + budget alert.
  2. Require prompt-injection check + secret-redaction.js scrub.
  3. Require failure-to-answer, topic relevancy, toxicity (+ thumbs).
  4. Require parented retrieve/llm/tool spans with evidenceIds.
  5. Refuse --clone-datadog.

Rubric

  • gold default → ok=true
  • empty {} or blind fixture → ok=false
  • --clone-datadog → datadog_clone_refused
  • doctor: npm run test:llm-obs-honesty PASS
  • evidence: command output in the same turn

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free