Add skill + exporter for dbt Cloud job completion DAG

[patkearns10]

Description

This PR adds a new dbt-extras Agent Skill that documents how to build a job-level DAG from dbt Cloud Admin API data—specifically job_completion_trigger_condition (run when another job completes)—which is separate from model lineage. It also ships a small Python exporter and a sample Mermaid file so the diagram can be opened in [Mermaid Live](https://mermaid.live/) or other renderers.

Motivation

Orchestration between jobs (including cross-project completion triggers) is easy to miss when looking only at model DAGs or schedules. The skill standardizes how agents gather jobs via dbtp ([dbt Rust SDK](https://github.com/dbt-labs/dbt-rust-sdk)), extract edges, and present results as JSON plus a styled flowchart LR diagram.

What’s included

• mapping-dbt-cloud-job-dag skill (SKILL.md): workflow for listing jobs, resolving project names, fetching full job definitions, building edges, sanitizing labels, and emitting Mermaid with ELK-friendly %%{init}%%, classDef / class styling, solid same-project links, and labeled dashed cross-project links.
• scripts/export_job_dag.py: optional CLI to generate a .mmd from dbtp (env: DBT_TOKEN, DBT_ACCOUNT_ID, DBT_HOST as needed); supports -cross-project-label, -renderer, and DBTP_PATH.
• job-dag-export.mmd: checked-in example output (regenerate with the script for your projects).
• Bump dbt-extras plugin version to 1.2.0 to reflect the new skill.

How to verify

1. Install dbtp per [dbt-rust-sdk](https://github.com/dbt-labs/dbt-rust-sdk) and configure DBT_* env vars.

2. Run (adjust project IDs / paths as needed):

   python3 skills/dbt-extras/skills/mapping-dbt-cloud-job-dag/scripts/export_job_dag.py
   --project-id <upstream_project_id> --project-id <downstream_project_id>
   -o /tmp/job-dag.mmd

3. Paste /tmp/job-dag.mmd into Mermaid Live and confirm subgraph colors, three completion edges, and the cross-project label render as expected.

Notes

• Edges reflect job completion triggers only; schedule/webhook ordering is out of scope (called out in the skill).
• job-dag-export.mmd is illustrative; consumers can delete it or replace it with a neutral fixture if the repo prefers not to ship account-flavored samples.

Checklist

• I have read the contributing guide and understand what's expected of me
• I have signed the CLA
• I have run this code in development and it appears to resolve the stated issue
• I have opened an issue to add/update docs, or docs changes are not required/relevant for this PR
• I have run changie new to create a changelog entry

[patkearns10]

Added to this:

• Progress logging — Writes interim status to stderr ([job-dag-export] prefix): scope summary, project/job counts after listing, milestones during bulk jobs get, connected-only before/after counts, render summary. Helps ~5+ minute account exports feel observable instead of silent.
• -quiet — Suppresses progress lines; output path on stderr at the end is unchanged.
• JSON snapshots — On each successful fetch-mode run, writes a .json next to the .mmd (same basename) with full jobs get payloads, project_names, capture metadata, and options used for the fetch. Snapshot is written after edge endpoints are resolved and before -connected-only, so the same file supports toggling connected-only offline.
• -no-snapshot — Skips writing the JSON when disk size or sensitivity matters.
• -from-snapshot <path> — Rebuilds edges from the snapshot and renders with -renderer, -cross-project-label, -connected-only, and o only—no dbtp or DBT_* required. Mutually exclusive with -account, -project-id, and -expand-downstream.
• Docs — SKILL.md updated for stderr progress, -quiet, and the snapshot / -from-snapshot workflow.

[patkearns10]

---- ---- ---- ----

[b-per]

The thing is dbtp is not public today.