list

List

The pipeline's final stage. The per-account stages each leave one artifact under .gtm/<slug>/; this stage reads across all of them and produces the two things a GTM team works from: a ranked CSV and a per-account dossier.

When to use

• A batch of accounts has been scored and you want the prioritized list.
• The user asks for "the target list", "the CSV", or "a dossier" for outreach.
• You're wrapping up a pipeline run and need the hand-off deliverables.

Inputs

Reads, per account dir under the artifact root:

• score.json (required — an account without it is skipped) for tier/score.
• enrich.json (optional) for company/domain.
• signals.json (optional) for detected intent signals + the hiring board.
• people.json (optional) for contacts or persona targets.

Workflow

1. Make sure the accounts you want included have been scored (score.json exists). Resolving people first enriches the list with contacts, but isn't required.

2. Run the build script:

   python3 ${CLAUDE_PLUGIN_ROOT}/skills/list/scripts/build_list.py
   

   • It scans every account dir under the artifact root, ranks accounts A → B → Nurture → Reject (and by score within a tier), and writes _report/accounts.csv + _report/dossier.md under the artifact root.
   • Reject-tier accounts are listed last and excluded by default; pass --include-reject to keep them in both outputs.

3. The script prints a one-line summary (accounts, actionable count, output paths, top 5). Point the user at the two files.

Output

• _report/accounts.csv — columns: rank, company, domain, tier, score, gates_failed, signals_found, hiring_provider, contacts, people_source, top_contact, top_contact_title, top_contact_email_status, top_contact_linkedin, rationale. The top contact is the highest-priority persona match for the account.
• _report/dossier.md — one section per account: tier/score, rationale, failed gates, score breakdown, detected intent signals (with an evidence snippet), the hiring board (provider/slug + how it was discovered), and the contacts — or the persona targets when no contacts were resolved.

Notes

• The dossier mirrors the scoring model so a rep can see why an account ranks where it does — gates, graded dimensions, and the signals that drove them — not just a number.
• Aggregation reads the artifact root directly, so it composes with entity fan-out: score N accounts in parallel, then build the list once over all of them.
• Stdlib-only and offline-testable: bash scripts/tests/test_list.sh.

This is the end of the pipeline: a prioritized, evidence-backed list to act on.