Purpose
Detect the faint, early-stage indicators of competitor strategy changes and market shifts — the weak signals that precede press releases and product launches — and convert raw inputs into a prioritized watchlist rated by confidence and urgency. Solves the problem of intelligence teams reacting to competitive moves only after they're public and irreversible.
When to Use
Invoke this skill when the user asks to:
- Scan competitor news, filings, or hiring data for early warning signs
- Build a competitive intelligence briefing from collected market inputs
- Identify what a competitor is likely planning before they announce it
- Spot an emerging market shift before it becomes mainstream
- Triage a pile of scattered signals into what actually matters now
- Establish an ongoing early-warning monitoring routine for a competitor set
Inputs Required
Before starting, collect:
- Competitor set: The specific companies or players being watched
- Signal inputs: The raw material to analyze — news items, job postings, patent/regulatory filings, exec moves, pricing changes, partnership announcements, social/forum chatter, analyst notes
- Market context: The category, where it is in its lifecycle, and current competitive dynamics
- Strategic concerns: What the user most fears or wants to anticipate (e.g., a new entrant, a price war, a product category expansion)
- Time horizon: How far out the user needs to see — next quarter, next year, multi-year
- Decision triggers: What internal actions a signal could prompt (defensive launch, repricing, M&A, repositioning)
If signal inputs are sparse, work the available data fully and flag which signal categories are blind spots that the user should instrument going forward.
Instructions
Step 1: Categorize Every Input by Signal Type
Sort each input into one of seven signal categories. The category determines how far ahead it forecasts.
- Hiring & talent signals (longest lead time, 6–18 months): Job postings, exec hires, team expansions. A burst of ML hires or a new VP of a category they don't currently serve precedes capability building.
- Capital & resource signals (6–12 months): Funding rounds, capex, facility leases, supply contracts. Money commits before product ships.
- IP & regulatory signals (3–18 months): Patent filings, trademark registrations, regulatory submissions, certifications. These leak intent with a fixed publication delay.
- Partnership & ecosystem signals (3–9 months): New integrations, channel deals, co-marketing, supplier shifts. Reveals where they're placing strategic bets.
- Product & pricing signals (1–6 months): Beta programs, feature deprecations, pricing page changes, packaging tests, geographic soft launches.
- Narrative & messaging signals (1–4 months): Shifts in website copy, conference talk topics, exec posts, analyst briefings. Companies test the language before the product.
- Behavioral & market signals (variable): Customer complaints clustering, churn rumors, forum sentiment, demand-side shifts that aren't yet visible in any competitor's actions.
Step 2: Apply Weak-Signal Detection Logic
For each categorized input, run it through four weak-signal tests:
- Anomaly check — Does this deviate from the competitor's established baseline behavior? A signal only matters relative to a norm.
- Convergence check — Do two or more independent signals point at the same conclusion? Convergence sharply raises confidence (a hiring signal + a patent filing + a messaging shift telling the same story).
- Directionality check — Is the trend accelerating, steady, or one-off? Acceleration upgrades urgency; isolated events get held as watch items.
- Disconfirmation check — What benign explanation exists? Discount signals that have an obvious non-strategic cause (routine backfill hire, seasonal pricing).
Step 3: Form Hypotheses
Cluster the surviving signals into 3–6 specific, testable competitive hypotheses. Each hypothesis states: what the competitor is likely doing, the signals supporting it, and the single signal that would confirm or kill it next.
State hypotheses as falsifiable claims, not vague worries: "Competitor X is preparing a mid-market product tier launching in Q3" — not "Competitor X may be moving downmarket."
Step 4: Rate Each Signal by Confidence and Urgency
Score every retained signal on two axes:
- Confidence (Low / Medium / High): Driven by source reliability, number of converging signals, and how directly the signal implies the conclusion. Single uncorroborated source = Low; multiple independent converging sources = High.
- Urgency (Watch / Act Soon / Act Now): Driven by lead time remaining before the move goes public and the cost of reacting late. Long-lead signals with time to prepare = Watch; signals indicating an imminent public move = Act Now.
Plot the result mentally on a 2×2: High confidence + Act Now is the alert tier; Low confidence + Watch is the monitoring tier.
Step 5: Recommend Response and Next Signals to Monitor
For each ranked signal or hypothesis:
- Recommended response: A specific internal action proportional to confidence and urgency — instrument deeper, brief leadership, accelerate a roadmap item, or hold and watch.
- Confirming signal to watch: The single next observable that would upgrade confidence — tells the user exactly where to point monitoring next.
Output Format
- Structure: Executive alert summary up top, then a ranked signal table, then hypotheses, then a monitoring plan
- Length: 800–1,100 words
- Format: Brief prose summary + a structured table for the ranked signals + bulleted hypotheses
- Sections: Top Alerts (1–3 highest confidence/urgency signals) → Ranked Signal Table (signal, type, confidence, urgency, what it implies) → Competitive Hypotheses (3–6, each with supporting signals and a kill/confirm signal) → Blind Spots (signal categories with no input) → Monitoring Plan (what to watch next and where)
- Tone: Intelligence analyst — sober, evidence-weighted, explicit about confidence levels; flags uncertainty rather than overstating
Quality Checklist
Before delivering output, verify:
- [ ] Every input is assigned to one of the seven signal categories
- [ ] Each retained signal carries both a confidence rating and an urgency rating
- [ ] At least one convergence relationship is identified (or its absence noted) — signals are not assessed in isolation
- [ ] Each hypothesis states the single signal that would confirm or kill it
- [ ] Disconfirming/benign explanations are addressed for medium- and high-confidence signals
- [ ] Blind spots are named — signal categories with no data the user should instrument
Notes
- The highest-value signals usually come from the longest-lead categories (hiring, IP, capital) — they buy the most preparation time but feel the least urgent, so they get ignored. Weight them deliberately.
- Convergence beats any single strong signal. One dramatic data point is often noise; three quiet ones pointing the same direction is the real warning.
- Resist the temptation to upgrade urgency to make the briefing feel important — a credible "Watch" is more useful than an inflated "Act Now" that erodes trust in the next report.
- If the same inputs are re-run periodically, track which hypotheses were confirmed or killed to calibrate the confidence scoring over time.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills