Tech Route Comparison
Provided by Patsnap Eureka.
Use this skill when the decision object is a route set rather than a company,
a market arena, or a concrete proposal package. The goal is to compare routes
on evidence, normalize the comparison basis, and end with a recommendation the
user can act on.
Use When
- The user asks which technical route is better, more mature, or lower risk
- Route comparison across two or more technical approaches
- Maturity, readiness, feasibility, or TRL evaluation
- Scenario-specific route recommendation (e.g., "which route for EV use?")
- Opportunity map or white-space scan across routes
- Management-ready technical pre-research report
Do Not Use
- The real question is one company in one topic → route to
company-tech-profile
- The real question is a multi-player arena or player map → route to
competitive-landscape
- The real question is a concrete proposal or initiation package → route to
rd-initiation-review
- The task is pure market sizing or commercial comparison with no technical core
- Legal patent opinion, FTO, or infringement analysis
Modes
| Mode |
When To Use |
Focus |
overview |
Broad technical landscape for a topic |
Route inventory, frontier scan, ecosystem structure |
compare |
Direct comparison of named routes |
Head-to-head comparison matrix, ranking, recommendation |
maturity |
Readiness or TRL assessment |
Maturity rubric, stage gates, deployment evidence |
opportunity |
White-space or opportunity scan |
Gap analysis, emerging routes, entry paths |
Default mode: compare when routes are named, overview when only a topic is given.
Core Principles
Scope Freeze Before Retrieval
Freeze these parameters before wide retrieval to prevent evidence drift:
topic: the technology domain
decision_question: what the user needs to decide
decision_use: directional route scan / go-no-go / diligence-grade
application_scenario: the specific use case or context (e.g., EV battery,
data center cooling, edge inference)
comparison_level: material-level / component-level / system-level
known_routes: user-provided or discovered route set
time_window: default last 3-5 years
Comparison Basis Must Be Explicit
Before ranking routes, write down:
- What routes are being compared (frozen definitions)
- What dimensions are used for comparison (performance, cost, maturity,
risk, scalability, etc.)
- What normalization rules apply (same application scenario, same
comparison level, same time window)
Do not compare shifting route buckets. Do not mix material-level evidence
with system-level evidence as if they were the same layer.
TRL And Maturity Require Explicit Rubric
Do not write TRL labels, readiness levels, or maturity claims without an
explicit rubric that defines what each level means in the current context.
A route with many weak signals is still weak — do not confuse volume with
confidence.
Counterevidence Is Required
For every route recommendation, actively search for and document:
- Evidence that contradicts the recommendation
- Conditions under which the recommendation would change
- Update triggers that should prompt re-evaluation
Tool Routing And Fallback
This skill works across multiple tool environments. Before retrieval, detect
which capabilities are available and select the highest tier that is reachable.
Tier 1 (Recommended): Structured Patent/Paper Retrieval
- Use the host's best structured patent and paper retrieval stack.
- This usually means fielded patent/paper search plus record-level deep fetch.
- Evidence grade: S/A
- Required capabilities: per-route search, keyword or semantic route discovery,
and deep-read of shortlisted records.
Tier 2 (Fallback): Web Research + Scholarly Companion Lane
- Use the host's best broad web research tool plus a scholarly companion lane
when available.
- Exa, Tavily, Brave, and domain-specific scholarly databases are good examples,
not hard requirements.
- Evidence grade: A/B
- Coverage loss vs Tier 1: route breadth and patent-structure claims are less precise.
Tier 3 (Fallback): Generic Web Search And Page Reading
- Use generic web search and page/PDF reading tools available in the host.
- Evidence grade: B/C
Tier 4 (Minimum): User Materials + Reasoned Synthesis
- No external tools required
- Evidence grade: C/U
Routing Rules
- Detect available capabilities at the start of the workflow.
- Select the highest available tier as the primary retrieval channel.
- When a tier is unavailable, explicitly state the downgrade.
- If the tool stack is degraded, lower confidence on route breadth and frontier claims.
Minimum Working Files
Create or update these files in a writable run folder:
request.md
workplan.md
method_decisions.md
comparison_basis.md
query_log.csv
source_index.csv
claim_ledger.csv
report.md
Recommended subfolders are described in references/workflow.md.
Default Workflow
Step 0: Freeze Scope
Confirm or infer the following before any retrieval:
topic, decision_question, decision_use
mode (overview / compare / maturity / opportunity)
application_scenario, comparison_level
known_routes (user-provided or to be discovered)
time_window, audience, deliverable
If the user did not specify routes, proceed with route discovery in Step 1.
If the user named routes, freeze them and proceed to Step 2.
If the topic is broad and not route-frozen yet, write a provisional route
taxonomy into comparison_basis.md before retrieval starts.
Step 1: Build Route Taxonomy (When Routes Not Given)
- Run 2-3 broad topic searches to discover candidate routes.
- Tier 1: structured patent/paper search around the topic
- Tier 2: targeted web research plus scholarly companion search
- Tier 3/4: generic web or user-provided materials
- Extract candidate routes from patent IPC clusters, paper themes, and
review articles.
- Freeze the route set in
comparison_basis.md before proceeding.
- If the route set is still too fuzzy, stop and narrow it with the user
instead of writing a superficial ranking.
Step 2: Freeze Comparison Basis
Write to comparison_basis.md:
- Route definitions (what each route means, technically)
- Comparison dimensions (performance, cost, maturity, risk, scalability,
deployment readiness, etc.)
- Normalization rules:
- Same application scenario for all routes
- Same comparison level (do not mix material-level with system-level)
- Same time window
- Scoring approach (if maturity or TRL mode):
- Explicit rubric with level definitions
- Evidence requirements per level
Step 3: Retrieve Evidence Per Route
For each route independently:
- Run 2-3 searches focused on the route's technical specifics.
- Tier 1: structured patent/paper search for the route
- Tier 2: targeted web research plus scholarly companion search
- Tier 3/4: generic web or user-provided materials
- Collect evidence across dimensions:
- Patent activity and key technical approaches
- Paper frontier and recent breakthroughs
- Standards, specifications, and regulatory status
- Product deployments, benchmarks, and commercial signals
- Deep-read 2-3 representative patents/papers per route for technical detail.
- Record all searches in
query_log.csv with per-route tagging.
Do not mix evidence across routes during collection. Keep per-route evidence
buckets separate until the normalization step.
Step 4: Normalize And Compare
- Build the comparison matrix: routes (rows) × dimensions (columns).
Fill each cell with evidence-backed assessments.
Example format:
| Dimension |
Route A |
Route B |
Route C |
| Performance |
High — [Patent: X] demonstrates Y |
Moderate — lab-scale only |
High — [Paper: Z] benchmark |
| Cost |
High — mature supply chain |
Unknown — no production data |
Low — expensive precursors |
| Maturity (TRL) |
TRL 7 — pilot production |
TRL 4 — lab validation |
TRL 5 — prototype |
| Risk |
Low — well-understood failure modes |
High — scaling unknowns |
Moderate — IP concentration |
Apply the scoring rubric (if maturity or TRL mode):
- Score each route per dimension using the explicit rubric
- Show the rubric alongside the scores
- Do not assign TRL or maturity labels without evidence
Run the counterevidence pass:
- For each route recommendation, search for contradicting evidence
- Document weakening conditions and update triggers
- If counterevidence is strong, adjust the recommendation
Separate every claim into:
- Verified fact: directly supported by patent, paper, or official source
- Evidence-backed inference: reasonable conclusion from multiple signals
- Open gap: insufficient evidence, state what is missing
Step 5: Synthesize And Draft
Write the report following the mode-appropriate output skeleton.
Output Skeletons
Overview Mode
- Executive summary: topic landscape and main route families
- Route taxonomy and definitions
- Per-route profile cards (2-3 paragraphs each)
- Frontier signals and emerging directions
- Key evidence references
Compare Mode
- Executive summary: which route is recommended and why
- Scope and comparison basis (explicit)
- Route-by-route profile cards
- Comparison matrix (routes × dimensions)
- Recommendation with conditions, tradeoffs, and update triggers
- Counterevidence and weakening conditions
- Key evidence references
Maturity Mode
- Executive summary: maturity landscape and readiness gaps
- Scope and maturity rubric (explicit)
- Per-route maturity assessment with evidence
- Maturity comparison matrix
- Stage-gate or readiness roadmap
- Risks and deployment blockers
- Key evidence references
Opportunity Mode
- Executive summary: where the opportunities are
- Route landscape and current coverage
- Gap analysis (sparse coverage + technical value + entry path)
- Emerging routes and early signals
- Recommended investigation priorities
- Key evidence references
Every claim must cite its source type and identifier, e.g., [Patent: CN1234567B],
[Paper: DOI or title], [Web: source name].
Completion Gates
All must pass before delivering the final answer:
Guardrails
- Do not confuse volume with confidence. A route with many weak signals is
still weak.
- Do not let vendor or company marketing outrank patents, papers, standards,
or directly inspectable technical evidence.
- Do not turn raw patent or paper counts into maturity claims without dedup
and scope caveats.
- Do not write TRL, readiness, or maturity labels without an explicit rubric.
- Do not compare material-level, component-level, and system-level evidence
as if they were the same layer.
- Do not force a strong conclusion when the user only gave a broad topic and
the evidence base is thin.
- Do not skip the counterevidence pass — every recommendation needs at least
one documented weakening condition.
- If paper or patent coverage is inadequate for a route, say so explicitly
before making strength claims.
- Do not present a route ranking as definitive when the comparison basis is
incomplete or the scoring rubric is missing.
- Keep the recommendation tied to the stated scenario — a route that is best
for one application may not be best for another.
Load These Files Only As Needed
- references/method-benchmark.md
- references/domain-playbooks.md
- references/workflow.md
- references/source-routing.md
- references/deliverables.md
- references/evidence-schema.md
- references/quality-gates.md
- templates/request-template.md
- templates/workplan-template.md
- templates/comparison-basis-template.md
- templates/report-outline.md
1---2copyright: "Copyright © Patsnap. All rights reserved."3name: tech-route-comparison4description:5 Evidence-backed comparison of two or more technical routes, architectures, or6 solution paths. Use when the user asks for technical pre-research, technical7 route comparison, route selection, TRL assessment, maturity assessment,8 readiness evaluation, opportunity mapping, or any management-grade technical9 report deliverable — even if they only ask for initiation analysis,10 committee-ready materials, feasibility, or "which route is better".11argument-hint: "[topic + route set + optional scenario/purpose]"12provider: "Patsnap Eureka"13compatibility: "Designed for Claude Code, Codex, and similar agent runtimes that can read/write local files and call whatever search tools are available."14deliverable-default: "Structured Markdown comparison report plus traceable evidence files; docx/pdf export is optional."15fallback-policy: "Prefer structured patent/paper retrieval when available; otherwise downgrade through domain-specific sources, Exa, Tavily, Brave, web, and a known-URL reader without blocking the run."16metadata:17 source: https://skillmd.com/skills/patsnap/tech-route-comparison18---1920# Tech Route Comparison2122Provided by Patsnap Eureka.2324Use this skill when the decision object is a route set rather than a company,25a market arena, or a concrete proposal package. The goal is to compare routes26on evidence, normalize the comparison basis, and end with a recommendation the27user can act on.2829## Use When3031- The user asks which technical route is better, more mature, or lower risk32- Route comparison across two or more technical approaches33- Maturity, readiness, feasibility, or TRL evaluation34- Scenario-specific route recommendation (e.g., "which route for EV use?")35- Opportunity map or white-space scan across routes36- Management-ready technical pre-research report3738## Do Not Use3940- The real question is one company in one topic → route to `company-tech-profile`41- The real question is a multi-player arena or player map → route to42 `competitive-landscape`43- The real question is a concrete proposal or initiation package → route to44 `rd-initiation-review`45- The task is pure market sizing or commercial comparison with no technical core46- Legal patent opinion, FTO, or infringement analysis4748## Modes4950| Mode | When To Use | Focus |51|------|------------|-------|52| `overview` | Broad technical landscape for a topic | Route inventory, frontier scan, ecosystem structure |53| `compare` | Direct comparison of named routes | Head-to-head comparison matrix, ranking, recommendation |54| `maturity` | Readiness or TRL assessment | Maturity rubric, stage gates, deployment evidence |55| `opportunity` | White-space or opportunity scan | Gap analysis, emerging routes, entry paths |5657Default mode: `compare` when routes are named, `overview` when only a topic is given.5859## Core Principles6061### Scope Freeze Before Retrieval6263Freeze these parameters before wide retrieval to prevent evidence drift:6465- `topic`: the technology domain66- `decision_question`: what the user needs to decide67- `decision_use`: directional route scan / go-no-go / diligence-grade68- `application_scenario`: the specific use case or context (e.g., EV battery,69 data center cooling, edge inference)70- `comparison_level`: material-level / component-level / system-level71- `known_routes`: user-provided or discovered route set72- `time_window`: default last 3-5 years7374### Comparison Basis Must Be Explicit7576Before ranking routes, write down:7778- What routes are being compared (frozen definitions)79- What dimensions are used for comparison (performance, cost, maturity,80 risk, scalability, etc.)81- What normalization rules apply (same application scenario, same82 comparison level, same time window)8384Do not compare shifting route buckets. Do not mix material-level evidence85with system-level evidence as if they were the same layer.8687### TRL And Maturity Require Explicit Rubric8889Do not write TRL labels, readiness levels, or maturity claims without an90explicit rubric that defines what each level means in the current context.91A route with many weak signals is still weak — do not confuse volume with92confidence.9394### Counterevidence Is Required9596For every route recommendation, actively search for and document:9798- Evidence that contradicts the recommendation99- Conditions under which the recommendation would change100- Update triggers that should prompt re-evaluation101102## Tool Routing And Fallback103104This skill works across multiple tool environments. Before retrieval, detect105which capabilities are available and select the highest tier that is reachable.106107### Tier 1 (Recommended): Structured Patent/Paper Retrieval108109- Use the host's best structured patent and paper retrieval stack.110- This usually means fielded patent/paper search plus record-level deep fetch.111- Evidence grade: S/A112- Required capabilities: per-route search, keyword or semantic route discovery,113 and deep-read of shortlisted records.114115### Tier 2 (Fallback): Web Research + Scholarly Companion Lane116117- Use the host's best broad web research tool plus a scholarly companion lane118 when available.119- Exa, Tavily, Brave, and domain-specific scholarly databases are good examples,120 not hard requirements.121- Evidence grade: A/B122- Coverage loss vs Tier 1: route breadth and patent-structure claims are less precise.123124### Tier 3 (Fallback): Generic Web Search And Page Reading125126- Use generic web search and page/PDF reading tools available in the host.127- Evidence grade: B/C128129### Tier 4 (Minimum): User Materials + Reasoned Synthesis130131- No external tools required132- Evidence grade: C/U133134### Routing Rules135136- Detect available capabilities at the start of the workflow.137- Select the highest available tier as the primary retrieval channel.138- When a tier is unavailable, explicitly state the downgrade.139- If the tool stack is degraded, lower confidence on route breadth and frontier claims.140141## Minimum Working Files142143Create or update these files in a writable run folder:144145- `request.md`146- `workplan.md`147- `method_decisions.md`148- `comparison_basis.md`149- `query_log.csv`150- `source_index.csv`151- `claim_ledger.csv`152- `report.md`153154Recommended subfolders are described in [references/workflow.md](references/workflow.md).155156## Default Workflow157158### Step 0: Freeze Scope159160Confirm or infer the following before any retrieval:161162- `topic`, `decision_question`, `decision_use`163- `mode` (overview / compare / maturity / opportunity)164- `application_scenario`, `comparison_level`165- `known_routes` (user-provided or to be discovered)166- `time_window`, `audience`, `deliverable`167168If the user did not specify routes, proceed with route discovery in Step 1.169If the user named routes, freeze them and proceed to Step 2.170171If the topic is broad and not route-frozen yet, write a provisional route172taxonomy into `comparison_basis.md` before retrieval starts.173174### Step 1: Build Route Taxonomy (When Routes Not Given)1751761. Run 2-3 broad topic searches to discover candidate routes.177 - Tier 1: structured patent/paper search around the topic178 - Tier 2: targeted web research plus scholarly companion search179 - Tier 3/4: generic web or user-provided materials1802. Extract candidate routes from patent IPC clusters, paper themes, and181 review articles.1823. Freeze the route set in `comparison_basis.md` before proceeding.1834. If the route set is still too fuzzy, stop and narrow it with the user184 instead of writing a superficial ranking.185186### Step 2: Freeze Comparison Basis187188Write to `comparison_basis.md`:189190- Route definitions (what each route means, technically)191- Comparison dimensions (performance, cost, maturity, risk, scalability,192 deployment readiness, etc.)193- Normalization rules:194 - Same application scenario for all routes195 - Same comparison level (do not mix material-level with system-level)196 - Same time window197- Scoring approach (if maturity or TRL mode):198 - Explicit rubric with level definitions199 - Evidence requirements per level200201### Step 3: Retrieve Evidence Per Route202203For each route independently:2042051. Run 2-3 searches focused on the route's technical specifics.206 - Tier 1: structured patent/paper search for the route207 - Tier 2: targeted web research plus scholarly companion search208 - Tier 3/4: generic web or user-provided materials2092. Collect evidence across dimensions:210 - Patent activity and key technical approaches211 - Paper frontier and recent breakthroughs212 - Standards, specifications, and regulatory status213 - Product deployments, benchmarks, and commercial signals2143. Deep-read 2-3 representative patents/papers per route for technical detail.2154. Record all searches in `query_log.csv` with per-route tagging.216217Do not mix evidence across routes during collection. Keep per-route evidence218buckets separate until the normalization step.219220### Step 4: Normalize And Compare2212221. Build the comparison matrix: routes (rows) × dimensions (columns).223 Fill each cell with evidence-backed assessments.224225Example format:226227| Dimension | Route A | Route B | Route C |228|-----------|---------|---------|---------|229| Performance | High — [Patent: X] demonstrates Y | Moderate — lab-scale only | High — [Paper: Z] benchmark |230| Cost | High — mature supply chain | Unknown — no production data | Low — expensive precursors |231| Maturity (TRL) | TRL 7 — pilot production | TRL 4 — lab validation | TRL 5 — prototype |232| Risk | Low — well-understood failure modes | High — scaling unknowns | Moderate — IP concentration |2332342. Apply the scoring rubric (if maturity or TRL mode):235 - Score each route per dimension using the explicit rubric236 - Show the rubric alongside the scores237 - Do not assign TRL or maturity labels without evidence2382393. Run the counterevidence pass:240 - For each route recommendation, search for contradicting evidence241 - Document weakening conditions and update triggers242 - If counterevidence is strong, adjust the recommendation2432444. Separate every claim into:245 - **Verified fact**: directly supported by patent, paper, or official source246 - **Evidence-backed inference**: reasonable conclusion from multiple signals247 - **Open gap**: insufficient evidence, state what is missing248249### Step 5: Synthesize And Draft250251Write the report following the mode-appropriate output skeleton.252253## Output Skeletons254255### Overview Mode2562571. Executive summary: topic landscape and main route families2582. Route taxonomy and definitions2593. Per-route profile cards (2-3 paragraphs each)2604. Frontier signals and emerging directions2615. Key evidence references262263### Compare Mode2642651. Executive summary: which route is recommended and why2662. Scope and comparison basis (explicit)2673. Route-by-route profile cards2684. Comparison matrix (routes × dimensions)2695. Recommendation with conditions, tradeoffs, and update triggers2706. Counterevidence and weakening conditions2717. Key evidence references272273### Maturity Mode2742751. Executive summary: maturity landscape and readiness gaps2762. Scope and maturity rubric (explicit)2773. Per-route maturity assessment with evidence2784. Maturity comparison matrix2795. Stage-gate or readiness roadmap2806. Risks and deployment blockers2817. Key evidence references282283### Opportunity Mode2842851. Executive summary: where the opportunities are2862. Route landscape and current coverage2873. Gap analysis (sparse coverage + technical value + entry path)2884. Emerging routes and early signals2895. Recommended investigation priorities2906. Key evidence references291292Every claim must cite its source type and identifier, e.g., `[Patent: CN1234567B]`,293`[Paper: DOI or title]`, `[Web: source name]`.294295## Completion Gates296297All must pass before delivering the final answer:298299- [ ] Route set is frozen before wide retrieval300- [ ] Comparison basis is explicit and written to `comparison_basis.md`301- [ ] Comparison level is consistent (not mixing material/component/system)302- [ ] At least 2 searches per route executed with results analyzed303- [ ] Per-route evidence collected independently before cross-route comparison304- [ ] Comparison matrix present with evidence backing per cell305- [ ] If TRL or maturity labels are used, explicit rubric is present306- [ ] Counterevidence pass completed for the main recommendation307- [ ] Every major claim has a traceable source citation308- [ ] Verified fact / evidence-backed inference / open gap clearly separated309- [ ] Tool tier and coverage limitations explicitly stated310- [ ] If evidence is insufficient for confident ranking, stated explicitly311 with next-evidence-gathering directions312313## Guardrails314315- Do not confuse volume with confidence. A route with many weak signals is316 still weak.317- Do not let vendor or company marketing outrank patents, papers, standards,318 or directly inspectable technical evidence.319- Do not turn raw patent or paper counts into maturity claims without dedup320 and scope caveats.321- Do not write TRL, readiness, or maturity labels without an explicit rubric.322- Do not compare material-level, component-level, and system-level evidence323 as if they were the same layer.324- Do not force a strong conclusion when the user only gave a broad topic and325 the evidence base is thin.326- Do not skip the counterevidence pass — every recommendation needs at least327 one documented weakening condition.328- If paper or patent coverage is inadequate for a route, say so explicitly329 before making strength claims.330- Do not present a route ranking as definitive when the comparison basis is331 incomplete or the scoring rubric is missing.332- Keep the recommendation tied to the stated scenario — a route that is best333 for one application may not be best for another.334335## Load These Files Only As Needed336337- [references/method-benchmark.md](references/method-benchmark.md)338- [references/domain-playbooks.md](references/domain-playbooks.md)339- [references/workflow.md](references/workflow.md)340- [references/source-routing.md](references/source-routing.md)341- [references/deliverables.md](references/deliverables.md)342- [references/evidence-schema.md](references/evidence-schema.md)343- [references/quality-gates.md](references/quality-gates.md)344- [templates/request-template.md](templates/request-template.md)345- [templates/workplan-template.md](templates/workplan-template.md)346- [templates/comparison-basis-template.md](templates/comparison-basis-template.md)347- [templates/report-outline.md](templates/report-outline.md)