estimatelisted

# Estimator — Estimate a Ticket Suggest a story-point estimate for a Jira ticket, grounded in the team's historical tickets, then record the outcome back into the dataset. The model does the fetch + analysis; the bundled `build_csv` script does the CSV write. ## Read-only Jira access This skill treats Jira purely as a **data source**. Use only read operations — `atlassianUserInfo`, `getAccessibleAtlassianResources`, project/field metadata, `searchJiraIssuesUsingJql`, `getJiraIssue`, `getJiraIssueRemoteIssueLinks`. You MUST NOT call any operation that creates, edits, transitions, comments on, links, assigns, deletes, or logs work against Jira issues (or writes to Confluence). If a step seems to require a write, stop and tell the user instead of proceeding. ## Data location Resolve `<DATA_DIR>` exactly as the `sync` skill does — the per-user data dir `~/.estimator`: - Bash: `DATA_DIR="$HOME/.estimator"` - PowerShell: `$DATA_DIR = Join-Path $env:USERPROFILE '.estimator'` Files: `<DATA_DIR>/history.csv` (the dataset) and `<DATA_DIR>/config.json` (holds `story_point_field_id` and `project_keys`, written by `sync`). ## Step 1 — Preflight 1. Resolve `<DATA_DIR>`. If `history.csv` does not exist, tell the user to run `/sync` first to build the baseline, and stop. If it exists but has no rows with `estimate_basis` of `actual` or `final`, warn that the suggestion will be low-confidence (little real history to compare against). 2. Verify the Atlassian MCP is connected