review-bounced-contacts
About
This skill automates a weekly manual review process for contacts with 3+ email bounces, helping developers decide whether to delete or attempt recovery for each contact. It prevents over-suppression by ensuring a human reviews auto-flagged data before permanent action. The process includes a scripted pre-filtering stage using the HubSpot API to prepare the review list.
Quick Install
Claude Code
Recommendednpx skills add TomGranot/hubspot-admin-skills -a claude-code/plugin add https://github.com/TomGranot/hubspot-admin-skillsgit clone https://github.com/TomGranot/hubspot-admin-skills.git ~/.claude/skills/review-bounced-contactsCopy and paste this command in Claude Code to install this skill
Documentation
Review Bounced Contacts
A weekly manual review process for contacts flagged with 3+ bounces. The bounce monitoring workflow auto-suppresses these contacts, but a human should decide whether to permanently delete or attempt recovery.
Prerequisites
- Bounce monitoring workflow active (run
/bounce-monitoring-workflowfirst) email_health_flagcustom property exists on contacts- HubSpot API token in
.env(for scripted pre-filtering)
Step-by-Step Instructions
Stage 1: Before — Pull the Review List
Use the HubSpot API to search for contacts where email_health_flag is set:
from hubspot import HubSpot
from hubspot.crm.contacts import PublicObjectSearchRequest
api_client = HubSpot(access_token=os.getenv("HUBSPOT_API_TOKEN"))
search = PublicObjectSearchRequest(
filter_groups=[{
"filters": [{
"propertyName": "email_health_flag",
"operator": "EQ",
"value": "true"
}]
}],
properties=["email", "firstname", "lastname", "company",
"hs_email_bounce", "hs_email_hard_bounce_reason_enum",
"lifecyclestage", "hubspot_owner_id"]
)
results = api_client.crm.contacts.search_api.do_search(search)
Export results to a CSV for review.
Stage 2: Execute — Review Each Contact
For each flagged contact, check:
- Is the email domain active? Run a quick MX record lookup or visit the domain.
- Is this a known customer or high-value contact? Check lifecycle stage and deal history.
- What is the bounce reason? Hard bounce (invalid mailbox) vs. soft bounce (mailbox full, temporary error).
Decision matrix:
| Domain active? | High value? | Bounce type | Action |
|---|---|---|---|
| No | Any | Any | Delete |
| Yes | No | Hard | Delete |
| Yes | No | Soft | Keep suppressed, recheck next quarter |
| Yes | Yes | Hard | Attempt to find updated email |
| Yes | Yes | Soft | Keep suppressed, monitor |
Stage 3: After — Execute Decisions
- Delete contacts marked for deletion via the HubSpot UI or API batch delete.
- Clear the
email_health_flagon all reviewed contacts. - Log the review results (deleted count, kept count, recovery attempts) for the quarterly report.
Stage 4: Rollback
- Deleted contacts can be restored from HubSpot's recycling bin within 90 days.
- Contacts kept as suppressed can be restored to marketing status via a workflow or manual update in the UI.
Frequency
Run weekly, ideally Monday morning. Should take 5-15 minutes depending on volume. If volume exceeds 50 contacts per week, investigate the root cause (bad list source, form spam, etc.).
GitHub Repository
Related Skills
llamaguard
OtherLlamaGuard is Meta's 7-8B parameter model for moderating LLM inputs and outputs across six safety categories like violence and hate speech. It offers 94-95% accuracy and can be deployed using vLLM, Hugging Face, or Amazon SageMaker. Use this skill to easily integrate content filtering and safety guardrails into your AI applications.
cost-optimization
OtherThis Claude Skill helps developers optimize cloud costs through resource rightsizing, tagging strategies, and spending analysis. It provides a framework for reducing cloud expenses and implementing cost governance across AWS, Azure, and GCP. Use it when you need to analyze infrastructure costs, right-size resources, or meet budget constraints.
quantizing-models-bitsandbytes
OtherThis skill quantizes LLMs to 8-bit or 4-bit precision using bitsandbytes, achieving 50-75% memory reduction with minimal accuracy loss. It's ideal for running larger models on limited GPU memory or accelerating inference, supporting formats like INT8, NF4, and FP4. The skill integrates with HuggingFace Transformers and enables QLoRA training and 8-bit optimizers.
dispatching-parallel-agents
OtherThis Claude Skill dispatches multiple agents to investigate and fix 3+ independent problems concurrently. It is designed for scenarios involving unrelated failures that can be resolved without shared state or dependencies. The core capability is parallel problem-solving, assigning one agent per independent problem domain to maximize efficiency.
