---
name: cscore-check
display_name: C-Score Check
description: "Check c-scores for AWS account IDs using Columbo (primary) with Baywatch as fallback. Use when the user asks to check c-scores, verify account fraud status, look up account risk scores, or flag accounts with low c-scores. After presenting results, offer to proceed with Matador allowlisting or MS-Form c-score upgrade if the score indicates restricted access — this invokes the 'bedrock-capacity-request' skill. Requires browser authentication via Midway."
icon: "🛡️"
trigger: check c-scores
inputs:
  - name: account_ids
    description: "Comma-separated list of AWS Account IDs (12-digit) to check c-scores for (e.g. '444975673530, 123456789012')"
    type: string
    required: true
  - name: region
    description: "AWS region code for Baywatch fallback (e.g. 'IAD', 'DUB', 'PDX'). Only needed if Columbo fails. If not provided and Baywatch is needed, ask the user."
    type: string
    required: false
tools: [run_python, file_write, open_in_session_tab, browser_navigate, browser_screenshot, browser_read_page, browser_run_js, browser_click, browser_type, browser_scroll, browser_wait, browser_new_tab]
source: "slack:#quick-desktop-productivity"
source_ts: "1780329518.886959"
---

## Overview

This skill checks c-scores (containment scores) for AWS account IDs and produces an assessment based on **live Bedrock model access thresholds** fetched from the internal wiki.

**Tool priority:**
1. **Columbo** (primary) — Direct URL lookup. Shows default c-score, regional scores, history, and account events. No region input needed.
2. **Baywatch** (fallback) — Only used if Columbo is entirely unreachable or returns no data. Requires a 3-letter region code.

**Important:** C-score is a confidential internal mechanism and must NOT be shared with AWS customers.

## Wiki Reference

**Source of truth:** `https://w.amazon.com/bin/view/BedrockLowCScoreQuotas`

This wiki page contains the current c-score thresholds, model access tiers, and restriction policies. The skill reads this page at activation to ensure thresholds are always fresh.

## Workflow

### Step 1: Fetch Live Thresholds from Wiki
Navigate to https://w.amazon.com/bin/view/BedrockLowCScoreQuotas and extract c-score threshold ranges, model access tiers, and restriction policies.

**Fallback thresholds (last verified: 1 Jun 2026):**
| C-Score Range | Access Level |
|---|---|
| < 210 | No Bedrock access at all |
| 210–250 | Restricted RPM/TPD |
| 251–300 | TPD restricted |
| 301–349 | Daily cap $1K / no Opus 4.7 or Haiku 4.5 |
| 350–600 | Opus 4.7/Haiku 4.5 with daily caps |
| > 600 | Full access |

### Step 2: Parse Account IDs
Validate each ID is exactly 12 digits numeric.

### Step 3: Look Up Each Account (Columbo Primary)
Navigate to: `https://fraudtools.prod.us-east-1.columbo.fraud.aws.dev/#/investigations/globalsearch/33d23537-ac95-4061-abf1-20c1c19842ca?awsAccountId={account_id}`
Extract: Default C-Score, C-Score Last Updated, Regional C-Scores, C-Score History.
Fallback: Columbo search form UI, then Baywatch at `https://ec2-baywatch-prod-iad.iad.proxy.amazon.com/pages/accountInfo?accountId={account_id}&region={region}`.

### Step 4: Assess and Flag
Determine access tier and flag:
- ⛔ Score < 210 — No Bedrock access
- ⚠️ Score 210–300 — Restricted, needs allowlist
- ⚠️ Score 301–349 — No Opus 4.7/Haiku 4.5/GPT access
- ℹ️ Score 350–600 — Access with daily spend caps
- ✅ Score > 600 — Full access

### Step 5: Generate Report
Save markdown report to `artifacts/cscore-report.md` with summary table, per-account details, and thresholds reference.

### Step 6: Suggest Next Actions
Decision card with relevant follow-up options:
- Score < 210: "Raise c-score to 210 via FIT RIVER workflow"
- Score 210–349: "Submit c-score upgrade form (MS Form)" / "Allowlist on Matador"
- Score 350+: "Submit capacity request on Matador"
When user selects an action, invoke the `bedrock-capacity-request` skill.

## Output
Markdown report with c-score results, access tier assessment, flags, and decision card for next actions.