Repository navigation
monitor: add complete SRE Agent visuals and VNet guidance - #46
Merged
Merged
Conversation
…tent Adds RED tests requiring the 16-file official asset set (12 SVGs + 4 PNGs), all 12 SVGs and 4 PNGs referenced in monitor/azure-sre-agent.md, and a new VNet connectivity section covering the three egress modes, subnet requirements, private DNS zones, ExpressRoute/VPN routing, preview limitations, and official source links. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…avior checks Task 1 review finding: the two preview-limitation assertions in test_briefing_covers_private_network_connectivity asserted exact, self-invented Korean sentences with no existing source to quote, coupling Task 2's prose to one specific phrasing instead of the underlying facts. Replace both with keyword/group checks over text windows following the '커넥터' and '아웃바운드'/'인바운드' anchors, so any equivalent natural Korean wording passes as long as it proves: 1. connector traffic uses the public internet and does not traverse the VNet during preview, and 2. VNet integration is outbound-only and inbound (private endpoint) is unsupported. Confirmed still RED: scripts/tests/ shows the same 6 failed, 62 passed as before this change (no regressions), and confirmed the new keyword checks independently pass against realistic equivalent Korean phrasing that differs from the previously hard-coded sentences. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…missions/memory diagrams Add 10 official Azure SRE Agent Learn media files (exact bytes preserved) to monitor/sre-agent-event-lab/assets/official/: observability diagnosis, notification paths, incident platform flow, memory knowledge sources, run modes comparison, permission flow, VNet networking screenshot, sub-agent canvas, managed connectors grid, and Operations Hub overview. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
… diagrams Add a new '## 비공개 네트워크에는 어떻게 연결하나요?' section covering VNet integration modes (Unrestricted/Limited/Azure VNet), subnet prerequisites (/28-/26, Microsoft.App/environments delegation, Network Contributor, SRE Agent Administrator), per-destination traffic routing, private DNS zone guidance, ExpressRoute/VPN reachability, Azure Policy governance, the separate Key Vault outbound-IP allow-list path, and a private-endpoint verification step. Place each newly downloaded official diagram/screenshot in its most relevant existing section (investigation sources, memory, notification behavior, incident routing, run modes/permissions, team extension, connectors, audit) with Korean descriptions and source links. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
… all 16 official assets Extends OFFICIAL_IMAGE_ALT_KEYWORDS from 6 to all 16 selected assets with image-specific keywords, requires every official asset to be rendered with Markdown image syntax (no raw HTML) and nonempty detailed Korean alt text, verifies path existence for every rendered image, forbids alt claims that are not visible in the artwork, forbids near-verbatim body duplication of alt text, and pins the official VNet details that are still missing. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…alt text and complete VNet guidance - Render all 16 selected official assets with Markdown image syntax, including incident-platform-flow.svg which was only a plain link before. - Replace raw HTML <img> fallbacks so alt text is verifiable by the test suite. - Correct the managed connectors alt text to the services actually shown (Jira, Slack, GitLab, Notion, Asana, Box, Confluence, OneDrive, SharePoint, GitHub, Azure DevOps, Office 365 Users, Viva Engage, Security Copilot, Freshdesk); Salesforce and Google Drive are not in the artwork. - Remove near-verbatim body sentences that duplicated the knowledge sources, notification paths, and sub-agent canvas alt text. - Expand the VNet screenshot alt text to cover the three egress modes, the delegated /28+ subnet, Microsoft.App/environments, private DNS, the connected state, the infra-network bypass warning, the MCP, package and code repository toggles, and additional hosts. - Add the official VNet details that were missing: additional hosts routing, preinstalled packages as an alternative to registry access, disconnecting the VNet before changing egress mode or subnet, and the required firewall domains including *.azuresre.ai with the Zscaler and WebSocket caveat. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…image alt accuracy Strengthen the briefing document contract so the final-review findings are caught by rules rather than by pinned phrases: - Add a general CommonMark emphasis lint that flags any delimiter run GitHub would render literally, including closing `**` glued to a Hangul particle, plus a self-check that the lint is not vacuous. - Detect duplicated substantive body sentences anywhere in the briefing and table rows whose description repeats the row label. - Require consistent private-network terminology. - Require the incident response flow alt text to follow the 0:00 to 5:00 timeline, including fix proposal or autonomous action, resolution, work item update, and team notification. - Require the Operations Hub alt text to describe the selected sidebar item and the empty chart state instead of implying plotted values. - Require every rendered image to carry Korean alt text that is not repeated as body prose. - Require the official network control rationale (data exfiltration, prompt injection), the transitional nature of infra network toggles, the auto-allowed hosts of configured packages and connectors, and the documented behavior when the network blocks a call. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
… VNet rationale Address the final-review findings on the briefing: - Move the closing bold marker so GitHub renders "VNet 통합" in bold instead of printing the asterisks, which CommonMark leaves literal when a closing `**` follows punctuation and is glued to a Hangul particle. - Describe the Operations Hub screenshot as it is: Operations Hub is the selected sidebar item and the chart area reports that no data exists for the selected period, rather than implying plotted values. - State that configured packages and connectors auto-allow their own hosts, so the additional host list only needs everything else. - Rewrite the incident response flow alt text to follow the diagram timeline from 0:00 to 5:00, including the fix proposal or autonomous action at 3:00 and the resolution with work item update and team notification at 5:00. - Remove the duplicated inbound limitation sentence and the repeated connector label while keeping both concepts in the traffic table and the limitations list. - Use 비공개 네트워크 and 프라이빗 엔드포인트 consistently in the new section. - Add the official rationale for network control (data exfiltration and prompt injection), the transitional nature of infra network toggles versus hostname-aware or FQDN egress filtering, the SRE Agent Administrator scope, and the documented behavior when the network blocks a call. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
Pull request overview
This PR updates the Azure SRE Agent guide and its verification tests by adding locally stored “official” visual assets (concept diagrams + selected portal screenshots) and expanding VNet integration guidance, with tests enforcing image/alt-text correctness and key networking content coverage.
Changes:
- Adds new official SVG diagrams and inserts them into
monitor/azure-sre-agent.mdwith detailed Korean alt text and sources. - Expands the VNet integration section with modes, prerequisites (subnet/RBAC), DNS/private endpoint notes, traffic routing, preview limitations, and firewall allow-list guidance.
- Strengthens pytest-based doc linting: emphasis rendering checks, duplicate prose detection, image rendering/path validation, and stricter alt-text assertions for official assets.
Reviewed changes
Copilot reviewed 2 out of 12 changed files in this pull request and generated 1 comment.
Show a summary per file
| File | Description |
|---|---|
| monitor/sre-agent-event-lab/scripts/tests/test_briefing_docs.py | Adds/extends doc lint tests for emphasis rendering, image/asset verification, alt-text validation, and expanded VNet content checks. |
| monitor/azure-sre-agent.md | Inserts official visuals, improves Korean alt text, and substantially expands VNet integration guidance and references. |
| monitor/sre-agent-event-lab/assets/official/run-modes-comparison.svg | New official diagram asset for review vs autonomous mode comparison. |
| monitor/sre-agent-event-lab/assets/official/permission-flow.svg | New official diagram asset describing managed identity vs OBO permission flow. |
| monitor/sre-agent-event-lab/assets/official/notification-paths.svg | New official diagram asset describing action/work-item/notification response paths. |
| monitor/sre-agent-event-lab/assets/official/knowledge-sources.svg | New official diagram asset summarizing uploaded docs vs MCP connector knowledge sources. |
| monitor/sre-agent-event-lab/assets/official/incident-platform-flow.svg | New official diagram asset for incident platform → agent → action → metrics flow. |
| monitor/sre-agent-event-lab/assets/official/diagnose-azure-services.svg | New official diagram asset for diagnostic data sources and agent correlation flow. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Fix review comment 3780214116: unrendered_emphasis_markers() treated CommonMark block-level `*` markers (list bullets like `* 항목`, and thematic breaks like `***`/`* * *`) as unclosable inline emphasis delimiters, since a lone `*` at a bullet position is neither left- nor right-flanking under the delimiter-run rules. Added test_emphasis_lint_ignores_block_level_star_markers, first captured RED against the unmodified lint, covering a valid Korean bullet, an indented bullet, an unindented bullet, and a thematic break. Fix: skip whole lines that are CommonMark thematic breaks, and skip individual `*` matches that are CommonMark list bullet markers (single `*`, <=3 spaces of indent, followed by whitespace or end of line) before running the flanking-delimiter scan. Genuine unclosable inline emphasis (including the Korean glued-particle case) is still caught; all previously passing assertions are unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
변경 사항
이미지 감사 결과
검증
pytest monitor/sre-agent-event-lab: 99 passed