Crewai Workflows

Execute AI-powered crew workflows for marketing content generation, customer support handling, data analysis, and social media calendar creation. Use when tasks involve (1) creating marketing content, taglines, or campaigns, (2) handling customer support inquiries or responses, (3) analyzing business data for insights, (4) generating comprehensive social media content calendars, or (5) any content generation or analysis task that benefits from specialized AI crew workflows. Workflows are powered by DeepSeek, Perplexity, and Gemini models.

Security Scan

The skill appears to do what it claims (call remote AI crews) but has several incoherencies and privacy risks: the runtime expects an API key and tools (curl/jq) that are not declared in the registry metadata, it includes a literal API key in the docs, and it sends potentially sensitive input to an unvetted third‑party server (crew.iclautomation.me).

Purpose & Capability

The skill's stated purpose is to run remote AI workflows — that matches the script and SKILL.md which call a remote server. However, the registry metadata claims no required env vars or binaries, while the runtime clearly expects an API key (CREWAI_API_KEY) and uses curl/jq. This mismatch is incoherent and could surprise users.

Instruction Scope

SKILL.md and scripts instruct the agent/user to POST arbitrary input (including support issues and business data) to https://crew.iclautomation.me. The skill therefore transmits whatever data the user provides to an external third party; there is no documentation about data handling, retention, or privacy. The docs also embed a literal API key example, which may be mistaken for a usable/default credential.

Install Mechanism

There is no install spec (instruction-only) which is low risk, but the included script requires external binaries (curl, jq) that the metadata didn't list. No remote downloads or archive extraction occur, which reduces supply-chain risk, but the script will fail or behave unexpectedly if expected tools are not present.

Credentials

The runtime requires an API key (CREWAI_API_KEY) to talk to the crew server, but the skill metadata does not declare any required env vars or a primary credential — an important omission. The SKILL.md shows a specific API key value inline; publishing a hardcoded key in docs is risky (could be a leaked/overly-permissive key or a decoy) and encourages users to use a third-party key without knowing scope/limits.

Persistence & Privilege

The skill is not always-enabled and does not request elevated persistence. The only file write is the script saving full responses to /tmp, which is limited in scope. Autonomous invocation (model invocation enabled) is default platform behavior and is not by itself a new risk here.

What to consider before installing

Before installing or using this skill: (1) Confirm who operates https://crew.iclautomation.me and read their privacy/retention policy — this skill sends whatever you pass (customer issues, business data) to that server. (2) Do not reuse the example API key printed in SKILL.md; treat it as untrusted. (3) Expect to need curl and jq on the host — the registry metadata should have declared those but didn't. (4) Avoid sending sensitive or PII data to this skill until you verify the third party's data handling and that the API key provided (if any) has appropriate scope and revocation options. (5) Ask the publisher for corrected metadata: explicitly declare CREWAI_API_KEY as a required credential and list required binaries, and provide a privacy/security statement for the crew server. If you cannot verify the operator or privacy practices, run the script in an isolated environment or prefer a vetted provider.

CrewAI Workflows Skill

Execute specialized AI crew workflows for content generation, analysis, and support tasks. All crews run on a dedicated server with production-grade LLMs.

Prerequisites

Set the API key as an environment variable (recommended):

export CREWAI_API_KEY="5aZyTFQJAAT03VPIII5zsIPcL8KTtdST"

Or pass it directly when calling the helper script.

Available Crews

1. Marketing Crew 📢

Generate marketing content, taglines, and campaign copy.

Use for:

• Product/service taglines
• Marketing copy for ads or landing pages
• Campaign messaging
• Value propositions

Input:

• topic (required) - What to create marketing content about
• target_audience (optional) - Who the content is for

LLM: DeepSeek
Response Time: 3-10 seconds

Example:

scripts/call_crew.sh marketing \
  '{"topic": "hypnotherapy for better sleep", "target_audience": "working professionals with insomnia"}'

---

2. Support Crew 🎧

Handle customer support inquiries with AI-generated responses.

Use for:

• Responding to customer questions
• Drafting support emails
• Handling common inquiries
• Escalation guidance

Input:

• issue (required) - The customer issue or question

LLM: DeepSeek
Response Time: 3-10 seconds

Example:

scripts/call_crew.sh support \
  '{"issue": "Client wants to reschedule their hypnotherapy session"}'

---

3. Analysis Crew 📊

Analyze business data and provide actionable insights.

Use for:

• Data interpretation
• Trend analysis
• Performance metrics review
• Business intelligence

Input:

• data_description (required) - Description of the data to analyze

LLM: DeepSeek
Response Time: 3-10 seconds

Example:

scripts/call_crew.sh analysis \
  '{"data_description": "Monthly client retention rates for Q4 2025"}'

---

4. Social Media Crew ⭐ 📱

Generate comprehensive 30-day social media content calendars with daily posts, captions, and hashtags.

Use for:

• Social media planning
• Content calendar creation
• Multi-platform content strategy
• Monthly content batches

Input:

• industry (required) - The business industry/niche
• company_name (required) - Business or personal brand name

LLMs: Perplexity (research) + Gemini (content generation)
Response Time: 3-5 minutes ⏳

Example:

scripts/call_crew.sh social_media \
  '{"industry": "hypnotherapy", "company_name": "Sidharth Mahto"}'

Note: This crew takes significantly longer due to comprehensive research and content generation phases.

---

Usage

Option 1: Using the Helper Script (Recommended)

cd crewai-workflows
scripts/call_crew.sh <crew_name> '<json_input>' [api_key]

Examples:

# Marketing crew
scripts/call_crew.sh marketing '{"topic": "sleep therapy for entrepreneurs", "target_audience": "startup founders"}'

# Support crew
scripts/call_crew.sh support '{"issue": "Client asking about session pricing"}'

# Analysis crew
scripts/call_crew.sh analysis '{"data_description": "Weekly session booking trends"}'

# Social media crew (takes 3-5 minutes)
scripts/call_crew.sh social_media '{"industry": "wellness coaching", "company_name": "Calm Mind Studio"}'

# With explicit API key
scripts/call_crew.sh marketing '{"topic": "mindfulness apps"}' "YOUR_API_KEY"

Option 2: Direct cURL

curl -X POST "https://crew.iclautomation.me/crews/<crew_name>/run" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $CREWAI_API_KEY" \
  -d '{"input": {...}}'

---

Response Format

All crews return structured JSON:

{
  "ok": true,
  "crew": "marketing",
  "trace_id": "abc123-def456",
  "result": {
    "workflow": "marketing",
    "output": "... the generated content ...",
    "input_summary": {...}
  },
  "error": null
}

Extract the output: The actual generated content is in result.output.

---

Best Practices

1. Set timeouts appropriately:

   • Marketing/Support/Analysis: 30-60 seconds
   • Social Media: 5-10 minutes

2. Check API key: Ensure CREWAI_API_KEY environment variable is set or pass explicitly

3. Handle errors: Check the error field in responses

4. Social Media crew: Expect 3-5 minutes response time; don't interrupt

5. Batch requests: For multiple similar tasks, consider running them sequentially

---

Health Check

Verify the CrewAI server is running:

curl https://crew.iclautomation.me/health
# Expected: {"ok": true}

---

Future Expansion

When new crews are added to the server:

1. Update the "Available Crews" section
2. Add example usage to the helper script
3. Document input parameters and response times

---

Server: https://crew.iclautomation.me
Authentication: API key via X-API-Key header
Last Updated: 2026-01-17