| name | competitive-research-snapshot |
| argument-hint | [company/product/segment, and the decision it supports] |
| description | Research a competitive landscape with cited snapshots, a comparison matrix, and so-what implications. Use when a product decision needs competitive grounding, not a market report. |
| intent | Autonomous competitive research as a workflow: search plan, competitor selection, just-enough research, Fact/Inference/Assumption labels, real URL citations, and next-step options — producing a stable snapshot schema that battle cards and delta monitors consume and diff. |
| type | workflow |
| theme | market-intelligence |
| best_for | ["Grounding a positioning, roadmap, or deal-support decision in cited competitive evidence","Creating the baseline snapshot that the competitive intel watch diffs against","Board or exec prep that needs labeled facts instead of confident storytelling"] |
| scenarios | ["We're rewriting our positioning next sprint — get me a cited read on our top three competitors","Sales keeps losing to one rival; I need evidence on where they're actually strong and weak"] |
| estimated_time | 20-40 min per run |
Competitive Research Snapshot
Purpose
Research a company's competitive landscape using a workflow, not a one-shot answer: search plan →
competitor selection → just-enough research → fact/inference labels → real URL citations → next-step
options. The output is a decision-support snapshot, not a market report — and because its schema is
stable, downstream skills (battle cards, delta monitors) can consume it and diff it. Because it
proceeds on labeled assumptions when questions go unanswered, it can run as an agent task or on a
schedule; re-run it and diff against the prior snapshot.
Input
Works best with: the company, product, or segment to research, and the decision this research
should support (positioning, roadmap bet, deal support, board prep) — the decision determines what
"just enough" means.
Also useful: known competitors (or explicit permission to identify them), and any prior snapshot
or market-landscape-scan output in session — the skill builds on evidence already gathered rather
than re-researching it.
Input supplied inline with the invocation — text after the skill name, a pasted context dump, or an
appended ARGUMENTS: line — counts as answers already given. Use it against the question budget;
don't re-ask.
Arriving empty-handed? That works too. The skill opens with at most 3 questions (subject,
decision, competitors) and proceeds on labeled assumptions if they go unanswered.
Example invocation: Competitive research snapshot on our expense-automation product — decision: which roadmap bet wins Q1. Competitors: [Competitor A], [Competitor B]; find a third if one matters.
Key Concepts
- Governing protocol: honors the
autonomous-investigation
contract in full — question budget of 3, search-plan gate, Fact/Inference/Assumption labels, Just
Enough Mode, stable schema, 4-option Final Step.
- Discipline mix: OSINT (press, reviews, analyst coverage) + FININT (filings, earnings language)
- Decision-support snapshot, not market report. Every section earns its place by serving the
decision named in Scope. Research value is decision support, not page count.
- Snapshot as baseline. The schema is a contract with future runs:
diffs the world against this document. Section order never changes.
