adversarial-research

Conduct comprehensive, cross-checked web investigations to verify claims and produce reliable reports.

no-skill baseline: 55% — anything below it makes the model worse.

1works
agent-ops-article-verificationskilldefault
Editorial teams or AI output validators needing pre-publication fact-checking with defense-in-depth claim extraction
2untested
loom-researchworkflow
Multi-angle research with parallel search, aggregation, and adversarial contradiction resolution.
3untested
mercury-multi-source-researchworkflow
Research questions needing ≥3 sources cross-checked for consistency (SDK/API behavior, vendor pricing, version facts); replaces serial autoresearch loop.
4untested
research-summarizerskill
Comparing multiple sources or extracting structured briefs beats manual reading.
5untested
fieldatlas-synthworkflow
Grounding research idea generation in corpus citations with novelty-checking and feasibility critique.
6untested
nihil-odinworkflow
When a transformation decision depends on external behavior, upstream docs, or evidence the repo cannot supply.
7untested
investigateworkflow
Orchestrate multi-agent investigate across parallel branches and synthesis.
8untested
deep-researchworkflow
Adversarial deep research with multi-source verification and ranked evidence synthesis.
9untested
ultraresearchskill
When you need maximum-saturation research across every available source.
10untested
researchworkflow
Complex multi-phase orchestration where deterministic ordering and structured outputs are critical.
11untested
researchplugin
Research queries that may benefit from specialist skills but need automatic routing to the right tool without requiring user to choose manually.
12untested
deep-researchcommand
When you need a deep, multi-source, fact-checked research report with confidence gates
13untested
deep-researchworkflow
When you need multi-phase orchestration with parallel agents across a complex workflow.
14untested
deep_researchworkflow
Executing complex multi-phase workflows with typed outputs.
15untested
rc-deep-researchworkflow
In-depth multi-source research when you need analyst-grade findings with proper citations.
16untested
researchworkflow
Exploratory research with parallel hypothesis agents and cross-check synthesis.
17untested
deep-researchworkflow
Producing a cited, high-confidence research report when sources include anti-bot paywalls (WebSearch aggregation fallback).
18untested
deep-researchskill
User wants thorough, cited research on any topic beating single web search with multi-source verification.
19untested
scientific-thinking-literature-reviewskill
Building citation-backed background sections where systematic protocol and deduplication rules prevent claim gaps.
20untested
kiln-researchworkflow
Deep research investigations with parallel multi-perspective evidence gathering.
21untested
pitch-atlas-craftsmen-researchworkflow
Coordinating multi-agent workflows with explicit phases, budget guards, and structured JSON output schemas.
22untested
researchworkflow
Fan-out web research, dedup sources, synthesize with citations, and fact-check claims.
23untested
mishkan-deep-researchworkflow
In-depth multi-angle research reports where contradictory sources must be reconciled by skeptic review.
24untested
research-flow-skillskill
Systematically researching a topic with explicit sourcing, contradiction detection, and confidence scoring for fact-checked reports
25untested
kmaz-researchworkflow
Adversarial research workflows that gather, refute, and converge on verified claims.
26untested
autoresearch-agentskill
When you need multi-source fact-checked research with explicit citations and adversarial verification.
27untested
annotate-partworkflow
multi-phase agent orchestration
28untested
research-engineskill
Rapid multi-source research synthesis with transparent evidence chains.
29untested
gemini-deep-researchskill
Multi-source synthesis tasks where systematic web search + AI analysis beats single-prompt research.
30untested
researchworkflow
Synthesizing evidence on academic/technical topics when you need multi-modality (text+code+data) sweep plus two-stage extraction-enrichment pipeline.
31untested
claim_verifierworkflow
Verify document claims are grounded in cited sources.
32untested
researchsubagent
When research requires multi-source adversarial verification and cited evidence trails.
33untested
cycle-cause-researchworkflow
multi-phase agent orchestration
34untested
note-correctness-reviewworkflow
multi-phase agent orchestration
35hurts
kb-articleskill
Support teams creating evergreen documentation from recurring customer questions or newly resolved issues.
36hurts
outreach-fanout-researchworkflow
When you need to build a prospect list from a look-alike pool of a known customer and you need both email AND phone verified from official sources, not brokers.