# Competitive Analysis > Competitive Analysis - Skill: `handsomeboy990/competitive-analysis` (Agent Skill, multi-file: 4 files) - Install (CLI): `npx skillmds@latest add handsomeboy990/competitive-analysis` - Raw SKILL.md: https://api.skillmd.com/api/skills/handsomeboy990/competitive-analysis/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Research & Search - Author: Handsomeboy990 (https://skillmd.com/u/handsomeboy990) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/handsomeboy990/competitive-analysis --- # Competitive Analysis A comparison is only as honest as its axes and its sources. The failure modes are choosing axes that flatter a predetermined winner, comparing options on different bases, and repeating each option's own marketing as if it were observed fact. This skill compares on evidence, for a decision, and states trade-offs rather than crowning a winner the decision did not ask for. ## 1. Define the axes from the decision ``` decision what the comparison is for; the axes serve it, not the other way axes the characteristics that actually change the decision, named before gathering, so the options cannot select the axes weight which axes are decisive and which are tie-breakers; a comparison that weights everything equally hides the one axis that matters exclude the axes that do not affect this decision, however tempting to list ``` Axes chosen after seeing the options tend to be the axes the preferred option wins. Choose them from the decision, first. ## 2. Gather each option's real characteristics ``` primary each option's real behaviour from primary sources: the docs, the pricing page, the source, the filing, a hands-on trial where possible not the option's own comparison page, which is engineered to win like for like the same axis measured the same way across options; a benchmark one vendor ran on its own hardware is not comparable to another's recency options change; a comparison against last year's version is stale ``` ## 3. Separate observed fact from claim ``` observed what was seen directly: the trial, the source, the measured number claimed what the option says about itself, attributed as a claim, not a fact independent what a third party without an interest reports, per source hierarchy gap where an option's real behaviour could not be established, say so; do not fill it with the option's marketing ``` ## 4. Compare and surface trade-offs ``` matrix options against axes, each cell an evidence-backed entry with its source and its type (observed, claimed, independent) tradeoff the axes on which options genuinely differ, and what each choice costs; the point of the analysis is the trade-off, not the score context the best option depends on the weight of the axes for this decision; state the conditions under which each option wins no false winner resist collapsing a real trade-off into a single ranking; a recommendation names the conditions it assumes ``` ## 5. The recommendation ``` conditional given the decision's weighting, this option, because of these axes alternatives when the weighting differs, which option wins instead risks what the recommendation depends on that could change gaps what could not be established and how it would move the answer ``` ## 6. Prohibitions - Never choose the comparison axes after seeing which option they favour. - Never compare options on different bases and present it as like-for-like. - Never repeat an option's marketing claim as an observed fact. - Never fill a gap in an option's real behaviour with its own marketing. - Never declare a single winner where a real trade-off exists; state the conditions. - Never weight every axis equally to avoid the judgement the decision needs. - Never trust an option's own comparison page as a source about a competitor. ## 7. Protocol 1. State the decision and derive the axes that affect it, with weights. 2. Gather each option's real characteristics from primary and independent sources. 3. Mark each cell as observed, claimed, or independent; name the gaps. 4. Compare like-for-like; assemble the evidence matrix. 5. Surface the genuine trade-offs and the conditions under which each option wins. 6. Recommend conditionally, with alternatives, risks, and gaps. 7. Verify the weighty claims with `source-verification` before relying on them. ## 8. Auto-critique Score from 0 to 5: axes derived from the decision before seeing the options, characteristics from primary and independent sources not marketing, observed kept distinct from claimed, like-for-like comparison, trade-offs surfaced rather than collapsed, recommendation conditional with its assumptions, gaps named. Threshold: no axis below 3, average at least 4. Axes chosen to flatter a predetermined winner, or marketing repeated as observed fact, caps the score until fixed. ## 9. Interfaces - Upstream: `research-core`, `source-research` gathers the evidence, `source-verification` checks the weighty claims. - Downstream: `synthesis-reporting` and `report-writing` render the analysis; `technology-selection` and `dependency-selection` are the engineering-specific forms of this comparison with their own criteria. - Lateral: `market-research` for the market-level version, `decision-records` when the choice is recorded. ## 10. Live sources The analysis reads each option's primary sources and independent reporting directly where the runtime can reach them. An option's own comparative marketing is read as a claim about itself, never as a source about a competitor. Gaps that require a source the runtime cannot reach are named, not filled.