From 59b3e3188f116b83745b4e78bc42f8bb6c975cb5 Mon Sep 17 00:00:00 2001
From: DB Lee <db.lee@microsoft.com>
Date: Thu, 14 May 2026 12:07:28 -0700
Subject: [PATCH] docs(end-to-end): correct init output tree and workflow file
 count

Re-validated docs/tutorial-end-to-end.md against current develop
(891 lines). Two drifts fixed.

- Section 3 'Initialize the workspace' claimed 'agentops init' produces
  '.agentops/results/' and '.github/skills/agentops-*/SKILL.md'. Reality
  (verified in /tmp/tut-135):
    * 'agentops init' creates exactly three files: 'agentops.yaml',
      '.agentops/data/smoke.jsonl', and a project-root '.gitignore'
      (only when one doesn't already exist).
    * '.agentops/results/' appears on the first 'agentops eval run', not
      at init time.
    * '.github/skills/...' is the output of 'agentops skills install',
      a separate command - not init.
  Replaced the tree with a 3-file list plus a clarifying paragraph.

- Section 7 'Generate the GitFlow workflows' said 'Four files appear
  under .github/workflows/'. 'agentops workflow generate' on current
  develop produces five files - the table was missing
  'agentops-watchdog.yml' (the daily cron that runs 'agentops doctor'
  against the run history). Added the fifth row.

Most other sections held up well:
  * Section 2 helper script ('scripts/create_support_agent.py') exists
    and ships the three tool definitions the table claims.
  * Section 5 evaluator inventory matches selector behaviour verified
    earlier (model-quality + agent evaluators with tool fields).
  * Section 9.1-9.4 watchdog wiring matches the real flow verified
    in #134 (azure_monitor / azure_resources / foundry_control all
    flip to 'ok' against real Azure resources).
  * Section 6 baseline comparison matches behaviour verified in #132.
  * Sections 8 OIDC / federated-credential setup is verbatim
    GitHub/Azure CLI, no AgentOps surface to drift.

Refs #135.
---
 docs/tutorial-end-to-end.md | 23 ++++++++++++-----------
 1 file changed, 12 insertions(+), 11 deletions(-)

diff --git a/docs/tutorial-end-to-end.md b/docs/tutorial-end-to-end.md
index e9aed20..1390013 100644
--- a/docs/tutorial-end-to-end.md
+++ b/docs/tutorial-end-to-end.md
@@ -188,18 +188,18 @@ In an empty folder (or the GitHub repo you want to use):
 agentops init
 ```
 
-You get:
+You get three files:
 
+```text
+agentops.yaml                  # at the project root
+.agentops/data/smoke.jsonl     # 3-row seed dataset
+.gitignore                     # only if one doesn't already exist
 ```
-agentops.yaml
-.agentops/
-├── data/
-│   └── smoke.jsonl
-└── results/
-.github/
-└── skills/
-    └── agentops-*/SKILL.md
-```
+
+`agentops init` does **not** create `.agentops/results/` (that appears on
+the first `agentops eval run`) or `.github/skills/`. Coding-agent skills
+live in their own command: `agentops skills install --platform copilot`
+(or `claude` / `cursor`).
 
 Open `agentops.yaml` at the project root and configure it for the
 support agent:
@@ -405,7 +405,7 @@ intentionally want to demonstrate a red quality gate.
 agentops workflow generate
 ```
 
-Four files appear under `.github/workflows/`:
+Five files appear under `.github/workflows/`:
 
 | Workflow | Trigger | Purpose |
 |---|---|---|
@@ -413,6 +413,7 @@ Four files appear under `.github/workflows/`:
 | `agentops-deploy-dev.yml` | Push to `develop` | Deploys to the **dev** environment after a passing eval. |
 | `agentops-deploy-qa.yml` | Push to a `release/*` branch | Deploys to **qa**. |
 | `agentops-deploy-prod.yml` | Push to `main` | Deploys to **prod** after a passing eval. |
+| `agentops-watchdog.yml` | Daily cron + `workflow_dispatch` | Runs `agentops doctor` against the run history and uploads the report as an artifact. |
 
 Read [`ci-github-actions.md`](ci-github-actions.md) for the full
 reference. The defaults are sane: you do not need to edit them yet.