/validate-profile

name: validate-profile
description: "Read-only health check that a brand profile is production-ready: required fields, voice and audience completeness, guardrails, compliance-jurisdiction coverage, connector configuration and MCP reachability, output-path writeability, and model-registry age — reported per check as BLOCKER or WARNING without ever printing credential values. Triggers on \"/digital-marketing-pro:validate-profile\", \"is the brand setup correct\", \"check connector credentials\", \"profile sanity check\", \"we rotated an API key — is it wired up\". Reads profile.json and probes connectors via connector-status.py; the prerequisite gate before /digital-marketing-pro:engagement, /digital-marketing-pro:campaign-plan, and /digital-marketing-pro:launch-campaign."
user-invocable: true
triggers:
  - validate brand profile
  - check brand profile health
  - is the brand setup correct
  - check connector credentials
  - profile sanity check
  - validate profile
allowed-tools: Read Bash Glob Grep

/digital-marketing-pro:validate-profile — Brand Profile + Credential Health Check

This skill is the canonical "is this brand ready to ship work?" gate. It validates a brand profile is complete enough for production use AND that every credential/connector referenced by the profile is actually reachable — **without ever printing credential values**.

Use this skill:

• After **`/digital-marketing-pro:brand-setup`** (or `/digital-marketing-pro:client-onboarding`) to confirm the new profile is production-ready.
• After **rotating any API key** (Slack, HubSpot, Stripe, Ahrefs, GA4 service account, etc.) so connectivity is re-confirmed without exposing the new value in logs.
• After **importing brand guidelines** (`/digital-marketing-pro:import-guidelines`) to confirm the merge succeeded.
• As the **prerequisite check** before `/digital-marketing-pro:engagement`, `/digital-marketing-pro:campaign-plan`, or `/digital-marketing-pro:launch-campaign`.

Why this skill exists

For agencies running 50–200 client brands, brand profiles and credentials drift constantly: a junior changes a Slack channel, an API key rotates, a brand voice gets edited. The cost of running a 60-minute engagement on a broken profile is hours of rework. This skill catches those drift cases in under 60 seconds.

The skill is **read-only** — it inspects state, never modifies it. It also **never prints credential values** — connector checks emit pass / fail / error-class without echoing the secret.

Validation dimensions

| Dimension | What's checked | Severity | |---|---|---| | **Required identity** | `brand_name`, `industry`, and a market/jurisdiction list non-empty — accept **`target_markets`** (what `brand-setup` actually writes) or `target_jurisdictions` (legacy). Checking only the legacy name made every freshly created brand fail its own validator on a BLOCKER | BLOCKER | | **Voice profile** | tone, formality and energy populated under **`brand_voice`** (what `brand-setup` writes, and what `content-engine` and `brand-voice-scorer.py` both read) or under `voice` (legacy). The generator is the source of truth here — two consumers already follow it, so this validator was the outlier | BLOCKER for content work | | **Audience profile** | `target_audience.primary_persona` with `role` + `reading_level` | WARNING | | **Guardrails** | `guardrails.prohibited_terms` + `guardrails.prohibited_claims` non-empty. `brand-setup` does not create this block, so report it as a WARNING with the exact command to add it for an unregulated brand, and reserve BLOCKER for regulated industries — where a missing guardrail is a real risk, not a setup gap | BLOCKER for regulated industries (pharma, BFSI, healthcare, legal); WARNING otherwise | | **Compliance jurisdictions** | Each declared jurisdiction has a matching rules entry in `skills/context-engine/compliance-rules.md` | BLOCKER | | **Connector config present** | Every connector named in `tracking.backend`, `integrations.*`, `analytics.*` has its env vars / `.mcp.json` entry present (local check; live reachability comes from the MCP/curl probe below) | BLOCKER per unconfigured connector | | **MCP server health** | Every entry in `.mcp.json` (if present) responds to a tools/list ping | WARNING | | **Credential storage** | `~/.claude-marketing/brands/{brand}/credentials.json` (or env vars) present for every backend referenced | BLOCKER | | **Output paths writeable** | `~/.claude-marketing/brands/{brand}/` is writeable; the user-visible publish dir (`$DIGITAL_MARKETING_PRO_PUBLISH_DIR` or `~/Documents/DigitalMarketingPro/`) is writeable | BLOCKER | | **Model curator currency** | `scripts/resolve_model.py --registry-age` returns < 90 days | WARNING |

A **BLOCKER** means "do not let the user run engagement / campaign-plan / launch-campaign until this is fixed." A **WARNING** is surfaced but does not gate.

Process

Step 0 — Resolve the brand to validate

If `--brand <slug>` was passed, use it. Otherwise read the active brand from `~/.claude-marketing/brands/_active-brand.json` (set by `/digital-marketing-pro:switch-brand`). If neither is available, error: `"--brand <slug> required, or run /digital-marketing-pro:switch-brand first."` Do NOT validate "everything" — validation is per-brand by design.

Step 1 — Load the brand profile

BRAND_DIR="$HOME/.claude-marketing/brands/{brand}"
test -d "$BRAND_DIR" || { echo "Brand directory not found at $BRAND_DIR — run /digital-marketing-pro:brand-setup first."; exit 1; }
PROFILE="$BRAND_DIR/profile.json"
test -f "$PROFILE" || { echo "profile.json missing under $BRAND_DIR — run /digital-marketing-pro:brand-setup."; exit 1; }

Parse the profile JSON and capture: `brand_name`, `industry`, `target_jurisdictions`, `voice.*`, `target_audience.*`, `guardrails.*`, `tracking.backend`, `integrations.*`, `analytics.*`.

Step 2 — Required-field checks

Walk the checklist in the **Validation dimensions** table above. For each field, r