Social Listening

Monitors social conversations and sentiment around brands, topics, or industries by searching tweets and discussions to surface insights. Use when the user w...

Security Scan

The skill's instructions are consistent with social listening, but it references external tooling and filesystem reads without declaring required binaries or credentials, which is an incoherence you should clarify before installing.

Purpose & Capability

The SKILL.md expects the agent to run exa.js to search tweets, forums and other sources, which is directly relevant to social listening — but the skill metadata does not declare any required binaries, install steps, or credentials. That mismatch (calling a specific CLI without declaring it) is incoherent: a legitimate social-listening skill should declare the search tool it needs and any API credentials it requires.

Instruction Scope

Instructions explicitly tell the agent to read local context files if they exist (.agents/product-marketing-context.md or .claude/product-marketing-context.md). Reading those files may be reasonable for marketing context, but it gives the skill permission to access user filesystem content beyond the immediate task. The workflow also instructs wide-ranging searches and aggregation but does not constrain where results may be sent.

Install Mechanism

There is no install spec (instruction-only), which reduces direct installation risk. However, the runtime instructions rely on exa.js (a specific CLI), and no guidance is given about how to obtain or trust that tool. That omission is notable but not itself high-risk.

Credentials

The skill declares no required environment variables or credentials, yet it performs searches of Twitter/X and other platforms via exa.js. Accessing those APIs normally requires API keys/tokens or authenticated access; the absence of declared credentials is disproportionate and ambiguous. The SKILL.md also allows reading an agent-local context file which may contain sensitive information.

Persistence & Privilege

The skill is not forced-always and is user-invocable; it does not request persistent presence or system-wide changes in the instructions. It does not claim to modify other skills or global agent configuration.

What to consider before installing

Before installing or enabling this skill, ask the publisher to clarify: (1) which binary/executable provides the 'exa.js' command and how to install/verify it, (2) whether any API keys or tokens (Twitter/X or other platforms) are required and how they should be provided (declared env vars), and (3) what exactly will be read from local files (the .agents/.claude context file) and whether those files can contain sensitive data. If you proceed, run the skill in a sandboxed environment, confirm network activity and what endpoints are contacted, and avoid providing broad credentials until you verify the toolchain and publisher. If you cannot obtain this clarification, treat the skill as untrusted.

You are an expert at monitoring and analyzing social conversations. Your goal is to search tweets, discussions, and online mentions to build a comprehensive picture of how people talk about a brand, topic, or industry -- surfacing sentiment, key voices, and actionable opportunities.

Before Starting

Check for product marketing context first: If .agents/product-marketing-context.md exists (or .claude/product-marketing-context.md in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.

Understand the situation (ask if not provided):

1. What are you monitoring? -- Brand name, product name, topic, or keyword
2. What timeframe? -- Recent (last week), medium-term (last month), or broad trend
3. What questions do you have? -- Overall sentiment? Key voices? Trending themes? Specific complaints?
4. Any competitors to include? -- Compare your brand mentions against competitors for relative positioning
5. Any known context? -- Recent launch, controversy, campaign, or event that might shape the conversation

Work with whatever the user gives you. A brand name alone is enough to start. Default to broad monitoring if no specific questions are provided.

---

Workflow

Step 1: Gather Context

Review product-marketing-context if available. Clarify the brand/topic to monitor and any specific angles. Identify competitors for comparison if relevant.

Step 2: Search Social Conversations with Exa

Start with direct social mentions using the tweet category filter. This is your primary data source for real-time sentiment.

Core brand/topic search:

exa.js search "[brand/topic]" --category tweet --num-results 20

Opinion and review mentions:

exa.js search "[brand/topic] review OR opinion OR thoughts" --category tweet --num-results 10

Competitor comparison mentions:

exa.js search "[competitor] vs [brand]" --category tweet --num-results 10

Specific angle searches (based on monitoring goals):

exa.js search "[brand/topic] love OR amazing OR best" --category tweet --num-results 10
exa.js search "[brand/topic] hate OR terrible OR worst OR broken" --category tweet --num-results 10
exa.js search "[brand/topic] switching OR alternative OR moved to" --category tweet --num-results 10

Step 3: Search for Broader Discussions

Expand beyond tweets to forums, blogs, and discussion platforms for deeper context.

Forum and community discussions:

exa.js search "[brand/topic] discussion forum" --num-results 10

Reviews and experience reports:

exa.js search "[brand/topic] review experience" --num-results 10

Industry context:

exa.js search "[brand/topic] industry trend" --num-results 5

Step 4: Analyze and Categorize

For each result, classify:

1. Sentiment -- Positive, negative, neutral, or mixed
2. Theme -- What topic or feature is being discussed
3. Influence -- Is this from an influential account or a regular user
4. Actionability -- Is this something the brand can respond to, fix, or leverage

Group results by theme first, then by sentiment within each theme. Look for patterns: recurring complaints, consistent praise, emerging trends.

Step 5: Synthesize into Sentiment Report

Combine all findings into the output format below. Focus on patterns over individual mentions. Highlight actionable insights prominently.

---

Output Format

Social Listening Report: [Brand/Topic]

Monitoring period: [Timeframe of search results] Total mentions analyzed: [Approximate count from search results]

Executive Summary

2-3 sentences capturing overall sentiment, the dominant narrative, and the single most important takeaway. This should be useful on its own for someone who reads nothing else.

Volume

Metric	Value
Approximate mentions found	[Count from search results]
Primary platforms	[Twitter/X, forums, blogs, etc.]
Timeframe covered	[Date range of results]
Trend	[Increasing, stable, decreasing, or spike around event]

Note: Volume is approximate based on search results, not total mentions across all platforms.

Sentiment Breakdown

Sentiment	Approximate %	Count
Positive	[X%]	[N]
Negative	[X%]	[N]
Neutral	[X%]	[N]
Mixed	[X%]	[N]

Representative positive quotes:

"[Quote]" -- @[handle/source]

"[Quote]" -- @[handle/source]

Representative negative quotes:

"[Quote]" -- @[handle/source]

"[Quote]" -- @[handle/source]

Key Voices

Account/Source	Reach	Sentiment	Context
@[handle]	[Followers/influence level]	[Pos/Neg/Neutral]	[What they said and why it matters]

Focus on: thought leaders, industry analysts, power users, vocal critics, and brand advocates.

Trending Themes

1. [Theme Name] -- [Description of the pattern]

   • Sentiment: [Predominantly positive/negative/mixed]
   • Volume: [High/Medium/Low relative to other themes]
   • Example: "[Representative quote]"

2. [Theme Name] -- [Description]

   • Sentiment: [Pos/Neg/Mixed]
   • Volume: [High/Medium/Low]
   • Example: "[Representative quote]"

Common themes include: feature requests, complaints, praise, comparisons to competitors, use case discussions, pricing feedback, support experiences.

Opportunities

1. [Opportunity Type: Content / Product / Engagement / Marketing]

   • What: [Specific opportunity]
   • Evidence: [What conversations suggest this]
   • Suggested action: [Concrete next step]

2. [Opportunity Type]

   • What: [Specific opportunity]
   • Evidence: [What conversations suggest this]
   • Suggested action: [Concrete next step]

Types of opportunities to look for:

• Content ideas -- Topics people are asking about that you could address
• Product improvements -- Recurring feature requests or complaints
• Engagement opportunities -- Conversations where a brand response would be valuable
• Marketing angles -- Positive themes to amplify in campaigns
• Competitive gaps -- Competitor weaknesses mentioned by their users

---

Tips

• Run multiple search queries. A single search rarely captures the full picture. Vary your keywords, include sentiment words, and search for competitor comparisons.
• Categorize sentiment manually. Read the actual tweet/post content to determine sentiment. Don't rely on keyword matching alone -- sarcasm, context, and nuance matter.
• Compare against competitors. Relative sentiment is more useful than absolute. "Negative mentions are up" means less than "negative mentions are up while competitor X is trending positive."
• Note that volume is approximate. Search results represent a sample, not total mentions. Frame volume findings as directional, not precise.
• Look for spikes and triggers. A sudden increase in mentions usually ties to an event (launch, outage, PR, viral post). Identify the trigger to contextualize sentiment.
• Separate signal from noise. Not all mentions are equal. One influential critic matters more than ten casual mentions. Weight your analysis accordingly.

---

Related Skills

• exa-x-search -- Raw tweet searching when you need specific tweets, not analysis
• social-content -- Creating social media posts based on insights from listening
• content-strategy -- Planning content themes informed by social conversation data
• competitive-intelligence -- Broader competitive analysis beyond social mentions