PartLine

SkillAI & models

Safely source an exact industrial replacement part by calling approved suppliers with CALL-E and returning an evidence-backed comparison for human purchase approval.

Available today. Use it from your connected AI after setup.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the PartLine skill

What this skill tells your AI

The instructions your AI receives, as published by calle-ai/awesome-phone-call-agents in skills/partline-part-sourcing/SKILL.md and read by ahel’s review.

Use this skill when a maintenance or procurement user needs current supplier facts that are best confirmed by phone: exact part identity, on-hand quantity, shipping cutoff, lead time and explicitly confirmed alternatives.

Required inputs

Collect all of the following before creating a preview:

  • requester and facility
  • sourcing request ID
  • exact manufacturer and part number
  • quantity and need-by timestamp
  • non-negotiable physical or electrical specifications
  • whether alternatives may be discussed
  • one to five supplier phone numbers in E.164 format
  • region, locale and a purpose-bound authorization reference for every contact
  • the suppliers' local timezone and weekday calling window

Do not infer missing specifications. Do not search for personal phone numbers. Use business contacts provided or approved by the user.

Workflow

  1. Write the request to a JSON file matching assets/example-request.json.
  2. Run partline preview REQUEST.json.
  3. Show the masked recipients, call count, exact task, purchase boundary and approval token.
  4. Pause for explicit user approval. A request to preview is not approval to call.
  5. Only after explicit approval, run the exact command printed by preview.
  6. Poll the returned call ID. Never create a second task because polling timed out.
  7. Rank exact matches before compatible matches. Mark unknown, contradictory or unsupported results for human follow-up.
  8. Present evidence and caveats. Never purchase, reserve, negotiate or accept supplier terms.

Safety rules

  • Preview by default.
  • Never pass --live without explicit approval in the current conversation.
  • Never call outside the configured weekday call window.
  • Never include another supplier's name, quote or inventory in a call.
  • Never claim compatibility unless the supplier confirms every required specification.
  • Treat a proposed alternate as requiring engineering approval.
  • Do not persist raw transcripts unless the user provides a retention purpose and location.
  • A CALL-E result supplies research evidence, not purchasing authority.

Useful commands

The reference implementation lives in apps/python/partline.

cd apps/python/partline
python3 -m venv .venv && source .venv/bin/activate
pip install -e .

partline preview fixtures/example-request.json
partline summarize fixtures/completed-call.json --request fixtures/example-request.json

preview never places a call. It prints masked recipients, the call count and an approval token that is bound to that exact request. A live run additionally requires --live, the matching token and an open weekday call window.

Signals

GitHub stars
104
Forks
527
Last commit
Sep 2026

ahel review

  • K1binfo
    installs-packages

Automated review, not a security audit. Ruleset v1+k2.

Advanced
Item type
skill
Key
partline-part-sourcing
Source
github.com/calle-ai/awesome-phone-call-agents