# Dbt Labs Docs Getdbt Com Docs Getdbt Com > Docs Issue Triage Skill - Skill: `tomevault-io/dbt-labs-docs-getdbt-com-docs-getdbt-com` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/dbt-labs-docs-getdbt-com-docs-getdbt-com` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/dbt-labs-docs-getdbt-com-docs-getdbt-com/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/dbt-labs-docs-getdbt-com-docs-getdbt-com --- # Docs Issue Triage Skill This skill reads open GitHub issues in `dbt-labs/docs.getdbt.com`, scores them by priority and size using the definitions below, presents a ranked list to the user, and — with their confirmation — applies the appropriate labels directly to GitHub. --- ## Step 1: Fetch open issues Use the `list_issues` tool to fetch open issues from `dbt-labs/docs.getdbt.com`. Supported options: - **Limit by count**: Use `perPage` to fetch a specific number (e.g. 10, 50, 100 max per request) - **Sort order**: Use `orderBy: CREATED_AT` + `direction: ASC` for oldest-first, `DESC` for newest-first - **Pagination**: Use the `endCursor` from `pageInfo` to fetch the next page For each issue, collect: - Issue number, title, and URL - Body / description - Existing labels (full list — needed for safe label merging in Step 4) - Comment count - Created date - Updated date **Skip** issues that already have both a `priority:` label AND a `size:` label — they've already been triaged. Track partially-labeled issues (priority but no size, or vice versa) separately — see Edge Cases. Before proceeding to Step 4, verify that all triage labels exist in the repo by checking the known label set: - `priority: high`, `priority: medium`, `priority: low` - `size: x-small`, `size: small`, `size: medium`, `size: large`, `size: x-large` If any label is missing from the repo, note it to the user before applying — do not create labels automatically. --- ## Step 2: Score each issue For each issue, assign one **priority label** and one **size label** based on the definitions below. Use the issue title, body, and any existing labels as signal. When in doubt, lean toward the more conservative (lower priority / smaller size) label and flag it for human review. ### Priority Labels | Label | When to apply | |---|---| | `priority: high` | Technical inaccuracy, missing or incorrect information, or broken links that negatively affect user workflows | | `priority: medium` | Fix or enhancement to existing information that is generating customer requests or confusion | | `priority: low` | Improvements that don't block workflows — grammar fixes, nice-to-haves, minor clarity improvements | **Priority signal heuristics:** - High comment count or many reactions → likely medium or high - Words like "wrong", "broken", "404", "incorrect", "missing" in title/body → likely high - Words like "typo", "grammar", "style", "suggestion" → likely low - Customer-facing language or mentions of support tickets → likely medium or high ### Size Labels | Label | When to apply | |---|---| | `size: x-small` | Under 3 hours to fix (e.g. fix a typo, update a single code snippet, fix a broken link) | | `size: small` | 1–2 days to address | | `size: medium` | Up to a week to address | | `size: large` | More than a week; may require more than one writer | | `size: x-large` | 1+ weeks of research for 1+ writers — e.g. information architecture projects, major restructuring | **Size signal heuristics:** - Single broken link or typo → x-small - One page needs updating → small - Multiple related pages or a new section needed → medium - New feature docs from scratch → large - IA changes, navigation restructuring, multi-page rewrites → x-large --- ## Step 3: Present ranked results Present the triaged issues as a ranked table, sorted by priority (high → medium → low), then by age (oldest first within each priority tier). Format: ``` ## Triage Results — [date] ### 🔴 Priority: High | # | Title | Suggested Size | Confidence | Notes | |---|---|---|---|---| | [#123](url) | Broken link in quickstart | x-small | High | 404 detected in title | ... ### 🟡 Priority: Medium ... ### 🟢 Priority: Low ... ### ⚠️ Partially Labeled (needs one dimension only) | # | Title | Has | Needs | Suggested | Confidence | Notes | |---|---|---|---|---|---|---| | [#456](url) | Some issue | `size: small` | priority | `priority: medium` | High | Support ticket origin | ... ``` - Link each issue number to its GitHub URL (e.g. `[#123](https://github.com/dbt-labs/docs.getdbt.com/issues/123)`) - Include a **Confidence** column (High / Medium / Low) to flag issues where the scoring is uncertain and human review is recommended before applying labels - List partially-labeled issues in their own section so they're clearly distinguished At the end of the table, ask: > "Would you like me to apply these labels to GitHub? I can apply all of them, only the high-confidence ones, or you can tell me which specific issues to label." --- ## Step 4: Apply labels (with confirmation only) Never apply labels without explicit user confirmation. Once confirmed: 1. For each confirmed issue, construct the full label list by **merging** the new triage labels into the issue's existing labels. Never pass only the new labels — this would strip all existing labels. - Example: existing labels `["content", "improvement", "paper cut"]` + new labels `["priority: medium", "size: small"]` → pass `["content", "improvement", "paper cut", "priority: medium", "size: small"]` 2. Use `issue_write` with `method: update` and the merged label list. 3. For partially-labeled issues, only add the missing dimension — preserve the existing triage label. 4. Report back with a summary of what was applied and flag any failures. **Label names must exactly match:** - `priority: high`, `priority: medium`, `priority: low` - `size: x-small`, `size: small`, `size: medium`, `size: large`, `size: x-large` --- ## Edge cases - **Already partially labeled** (has priority but not size, or vice versa): Score the missing dimension only, show in the dedicated partially-labeled section, and only apply the missing label. - **Ambiguous issues**: If the issue is too vague to score confidently, assign a Low confidence rating and suggest the user review before applying. - **Large batch**: If there are more than 50 un-triaged issues, offer to triage in batches by priority tier or by date range. --- > Source: [dbt-labs/docs.getdbt.com](https://github.com/dbt-labs/docs.getdbt.com) — distributed by [TomeVault](https://tomevault.io).