Forge and standardize public TSG templates - #342
Conversation
Align all authoring templates with safety, applicability, validation, persona, and automation requirements. Add the canonical public TSG metadata schema and provenance vocabulary. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
There was a problem hiding this comment.
Pull request overview
Standardizes Azure Local supportability authoring with safety, validation, applicability, and automation guidance.
Changes:
- Expands article and component templates with consistent authoring prompts.
- Adds the canonical TSG metadata schema.
- Updates contribution and review guidance.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
TSG/Templates/tsg-metadata.schema.json |
Defines metadata validation. |
TSG/Templates/Troubleshoot-Template.md |
Expands troubleshooting guidance. |
TSG/Templates/Reference-Template.md |
Standardizes reference articles. |
TSG/Templates/Overview-Template.md |
Adds routing and validation guidance. |
TSG/Templates/Markdown-Snippets.md |
Adds reusable authoring patterns. |
TSG/Templates/HowTo-Template.md |
Expands procedural safety guidance. |
TSG/Templates/DeepDive-Template.md |
Adds evidence and operational boundaries. |
TSG/Templates/Component/README-Template.md |
Standardizes component inventories. |
TSG/Templates/Component/CONTRIBUTING-Template.md |
Defines component contribution requirements. |
CONTRIBUTING.md |
Updates repository authoring requirements. |
.github/copilot-instructions.md |
Updates automated review guidance. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Address five template-consistency and correctness findings: 1. Reference-Template source-of-truth column no longer permits OEM documentation for the OEM row; it requires a public Microsoft source, matching the template's own public-Microsoft-only link policy. 2. Markdown-Snippets validation-record lint result now lists A, B, C, F (C was omitted though the schema and rubric allow it). 3. CONTRIBUTING.md grade table now uses the canonical type-completeness rubric from README-Template.md instead of detector/live-loop definitions; validation depth stays in fidelity_level. 4. .github/copilot-instructions.md grade table aligned to the same canonical rubric for the same reason. 5. Troubleshoot-Template rollback and rollback verification are split into a separate, explicitly conditional block so copying the mitigation block no longer runs rollback after a successful action.
The fidelity_level note I added said (L0 to L3), but the fidelity tables in both files, the canonical README-Template rows, and the schema enum define L4 (full live inject-detect-mitigate loop). Correct the range to L0 to L4 and reword the parenthetical so it spans static-only through the full live loop.
|
Reviewed the whole PR and read every changed file end to end, then ran the checks that matter for a template rewrite instead of just reading the diff. What held up:
The risk-label vocabulary is applied consistently and the read-only action type is kept separate from the state-changing risk labels. No issues to flag. |
T.J. Hambley (tihamble)
left a comment
There was a problem hiding this comment.
Reviewed all 11 files end to end, audited every deleted line against base, and validated the five article metadata markers against the new schema. No substantive guidance was dropped and nothing was renamed or removed. Approving.
Summary
azure-local-supportability/tsg-metadata/v1JSON Schema.Validation