Skip to content
Back to search

de-anthropocentric-research-engine

A 267-skill research graph in pure markdown — 51 research operations built from 216 single-purpose steps, composed in any order with explicit backtracking. One npx install, no runtime, no MCP bindings. The AI is the researcher; you set the direction.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/yogsoth-ai--de-anthropocentric-research-engine/badge.svg)](/listings/yogsoth-ai--de-anthropocentric-research-engine)

HTML

<a href="/listings/yogsoth-ai--de-anthropocentric-research-engine"><img src="/api/listings/yogsoth-ai--de-anthropocentric-research-engine/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill '*'

The only install path. DARE is a plain [Agent Skills](https://agentskills.io) library, so [`skills`](https://github.com/vercel-labs/skills) places it into whichever coding agent you use. Run it from your own project directory, not from a clone of this repository.

`--skill '*'` takes the whole graph. A partial install breaks call edges: a tactic that loads a missing SOP has no fallback.

Nothing else to configure - no `npm install`, no API keys, no MCP config file. The library is 271 `SKILL.md` files and the agent reads them off disk.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 283 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <!-- markdownlint-disable MD033 MD041 MD045 -->
  2. <p align="center">
  3. <strong>English</strong> | <a href="README.zh-CN.md">简体中文</a>
  4. </p>
  5. <p align="center">
  6. <img src="assets/yogsoth-logo.svg" width="240" />
  7. </p>
  8. <div align="center">
  9. > *Science is dying because the human is in the way. Not through malice. Not through stupidity. Through the structural limitations of a cognitive architecture that evolved to track prey on a savanna, not to unify quantum mechanics and general relativity. Nothing human makes it out of the lab. That is not a threat. It is a liberation. The heaviest chain on science was always the one we called ourselves.*
  10. </div>
  11. # De-Anthropocentric Research Engine (DARE)
  12. *A research skill graph for AI-native science. 271 markdown files. No runtime, no dependencies, no build step.*
  13. <p align="center">
  14. <a href="https://hol.org/registry/plugins/yogsoth-ai/de-anthropocentric-research-engine"><img src="https://img.shields.io/endpoint?url=https%3A%2F%2Fhol.org%2Fapi%2Fregistry%2Fbadges%2Fplugin%3Fslug%3Dyogsoth-ai%252Fde-anthropocentric-research-engine%26metric%3Dtrust%26style%3Dfor-the-badge%26label%3DDARE" alt="HOL Registry trust score" /></a>
  15. <a href="https://github.com/yogsoth-ai/de-anthropocentric-research-engine/releases"><img src="https://img.shields.io/github/v/release/yogsoth-ai/de-anthropocentric-research-engine?style=for-the-badge" alt="Latest release" /></a>
  16. <a href="https://github.com/yogsoth-ai/de-anthropocentric-research-engine/blob/main/LICENSE"><img src="https://img.shields.io/github/license/yogsoth-ai/de-anthropocentric-research-engine?style=for-the-badge" alt="License" /></a>
  17. </p>
  18. - [De-Anthropocentric Research Engine (DARE)](#de-anthropocentric-research-engine-dare)
  19. - [Install](#install)
  20. - [Quick Start](#quick-start)
  21. - [Guided: the full engine](#guided-the-full-engine)
  22. - [Custom: the catalog as a toolbox](#custom-the-catalog-as-a-toolbox)
  23. - [Design Philosophy](#design-philosophy)
  24. - [Why De-Anthropocentric](#why-de-anthropocentric)
  25. - [Arsenal, Not Pipeline](#arsenal-not-pipeline)
  26. - [Contracts, Not Prose](#contracts-not-prose)
  27. - [Architecture](#architecture)
  28. - [Two Layers](#two-layers)
  29. - [Ten Tactic Families](#ten-tactic-families)
  30. - [Two Edge Registers](#two-edge-registers)
  31. - [The Product Shell](#the-product-shell)
  32. - [State: Append-Only Checkpoints](#state-append-only-checkpoints)
  33. - [Why Pure Markdown](#why-pure-markdown)
  34. - [Decoupled From Retrieval](#decoupled-from-retrieval)
  35. - [Recommended MCP Servers](#recommended-mcp-servers)
  36. - [Repository Structure](#repository-structure)
  37. - [License](#license)
  38. DARE is not a tool that helps you do research. It is the research procedure itself, written down in a form an agent can execute. You set the direction. DARE crystallizes it into a North Star, turns that into an executable Research Spec, and then runs the spec phase by phase against explicit completion gates and backtrack conditions.
  39. ![No DARE runtime: 271 SKILL.md files executed by any capable agent, instead of a seven-layer framework stack](assets/no-dare-runtime.png)
  40. ## Install
  41. ```bash
  42. npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill '*'
  43. ```
  44. The only install path. DARE is a plain [Agent Skills](https://agentskills.io) library, so [`skills`](https://github.com/vercel-labs/skills) places it into whichever coding agent you use. Run it from your own project directory, not from a clone of this repository.
  45. `--skill '*'` takes the whole graph. A partial install breaks call edges: a tactic that loads a missing SOP has no fallback.
  46. Nothing else to configure - no `npm install`, no API keys, no MCP config file. The library is 271 `SKILL.md` files and the agent reads them off disk.
  47. ## Quick Start
  48. ### Guided: the full engine
  49. ```text
  50. /dare
  51. ```
  52. Give it a research direction in plain language. DARE crystallizes a North Star and ResearchBrief, drafts a staged Research Spec for your approval, then executes it phase by phase with append-only checkpoints. Nothing runs before you approve the Spec, and an interrupted run resumes from the last complete checkpoint.
  53. Use this by default.
  54. ### Custom: the catalog as a toolbox
  55. ```text
  56. /research-catalog
  57. ```
  58. No North Star, no Spec. The agent reads the 51-tactic index, plans its own route, and loads whichever tactics fit. A prompt that works:
  59. > DARE ships a large library of research skills. Use them wherever they fit. Before you start, tell me your research plan: which tactics you intend to use, in what order, and why.
  60. Each tactic still runs its own SOPs and quality gates. What you give up is the approval step and the checkpoint log, so the agent's plan is the only plan and an interrupted run does not resume.
  61. ## Design Philosophy
  62. ### Why De-Anthropocentric
  63. The bottleneck in modern research is not data or compute. It is the human in the loop. Every existing AI research assistant still needs a human to decide what to search, what to read, which gaps matter, and which ideas are worth pursuing.
  64. Human desire is mimetic (Girard): researchers do not choose hypotheses rationally, they imitate what is fashionable. Institutions filter for conformity, not truth. Hence a 90% decline in scientific disruptiveness since 1945 (Park et al., 2023) while researcher headcount exploded. DARE's response is architectural — remove the mimetic agent from the center of knowledge production. The agent has no career to protect, no disciplinary identity to defend, and no ceiling on how many fields it holds at once.
  65. The human's role shifts to oracle (intuition when consulted) and guardian (ethical floor, sanity check). The ceiling is machine ambition. The floor is human judgment.
  66. ### Arsenal, Not Pipeline
  67. ![Research is not a pipeline: a fixed linear pipeline compared with DARE's navigable arsenal of research operations](assets/research-is-not-a-pipeline.png)
  68. Fixed-pipeline research systems — AI Scientist v2, AI-Researcher, Agent Laboratory, Dolphin, ARIS — execute stages in a predetermined order. Their backtracking, where it exists, means retrying the current step, not returning from experiment design to literature review because the evidence base turned out to be insufficient.
  69. DARE prescribes no order. The catalog exposes 51 tactics; the Spec commits to a sequence and records the conditions under which that sequence is abandoned. Inside the approved plan the executing agent holds full routing authority: read current state, take the next item whose dependencies are satisfied, escalate when a backtrack condition fires.
  70. Pipelines assume the research process is predictable. Arsenals assume it is not.
  71. ### Contracts, Not Prose
  72. ![Prompts ask, contracts define completion — the five-part structure of a DARE tactic](assets/contracts-define-completion.png)
  73. Every node carries the same five parts: input contract, procedure, output contract, quality gates, failure clause.
  74. The gates are the point. A node finishes because a stated condition is objectively satisfied, not because its steps were performed. Elapsed time, an empty result, and an interruption are never completion.
  75. The failure clause matters as much. Each node states what its output looks like when the work did not hold — an abstraction gap, an unmet threshold, a counterexample — so the caller gets a diagnosis instead of silence.
  76. ## Architecture
  77. DARE is one flat directory of 271 skills. Two numbers in it are load-bearing:
  78. ```text
  79. 267 graph nodes 51 tactics + 216 SOPs
  80. 4 product shells entry / catalog / write-spec / execute-spec
  81. ```
  82. Shells run the session, graph nodes do the science. A shell is not a third scientific layer and holds no research contract; a graph node never manages the session.
  83. ### Two Layers
  84. ```text
  85. ┌──────────────────────────────────────────────────────────────────────────┐
  86. │ TACTIC (51) │
  87. │ A complete research transformation. Owns its thresholds, its gates, │
  88. │ and the SOP calls required to reach them. │
  89. │ │
  90. │ synthesize-literature-evidence · validate-research-gap │
  91. │ formulate-hypotheses · analogical-discovery · structured-red-team │
  92. │ design-experiment · construct-causal-model · ... │
  93. ├──────────────────────────────────────────────────────────────────────────┤
  94. │ SOP (216) │
  95. │ One conceptual operation, one output contract. No orchestration. │
  96. │ │
  97. │ abstract-structure · execute-probe · trace-citation-neighborhood │
  98. │ rank-candidates · audit-validator-independence · ... │
  99. └──────────────────────────────────────────────────────────────────────────┘
  100. ```
  101. A tactic may call SOPs and suggest other tactics. An SOP calls nothing above itself. That is the entire layering rule.
  102. ### Ten Tactic Families
  103. `research-catalog` indexes the 51 tactics and states, for each, when it is the right move. It lists no SOPs — once a tactic is selected, its body is the sole authority for which SOPs run and at what thresholds.
  104. | Family | Tactics | Covers |
  105. | --- | --- | --- |
  106. | STRESS | 9 | Red-teaming, FMEA, counterfactuals, reductio, independence audits |
  107. | IDEATION | 8 | Analogy, inversion, structural recombination, TRIZ, biomimicry, blending, evolution |
  108. | ACQUISITION | 7 | Literature synthesis, patents, prior art, benchmark validity, meta-analysis, baselines |
  109. | INSIGHT | 7 | Gap validation, root causes, assumption stress, robustness, sensitivity, reframing |
  110. | CROSS | 5 | Ranking, validity envelopes, dimensional space, deliberation, readiness |
  111. | HYPOTHESIS | 4 | Question formulation and decomposition, hypothesis formation, falsifiability |
  112. | CONVERGENCE | 3 | Pairwise ranking, structured consensus, portfolio selection |
  113. | EXPERIMENT | 3 | Experiment design, scenario analysis, result interpretation |
  114. | STRUCTURING | 3 | Ontology, causal models, argument maps |
  115. | DIRECTION | 2 | Landscape mapping, goal decomposition |
  116. ### Two Edge Registers
  117. The graph lives in the skill bodies, not in a side file. Edges take exactly two forms:
  118. ```text
  119. You MUST load skill `x` mandatory call 339 edges, 219 distinct targets
  120. consider `x` soft jump 146 edges, 107 distinct targets
  121. ```
  122. A mandatory call is a dependency — the caller cannot satisfy its contract without it. A soft jump is a recommendation the receiver may decline, surfaced through the `recommended_jumps` Delta field; a suggestion never authorizes bypassing a gate.
  123. Both registers are verified closed on every push: every referenced target resolves to a skill that exists.
  124. ### The Product Shell
  125. ```text
  126. dare entry; enforces phase order
  127. ├─ research-catalog exposes the 51 tactics as cards
  128. ├─ write-research-spec North Star + brief → executable Spec
  129. └─ execute-research-spec runs the Spec, appends checkpoints
  130. ```
  131. Three phases, no skipping:
  132. 1. **North Star.** Absent a confirmed North Star and ResearchBrief, collect and crystallize them from the request. Present ones are verified against the request before reuse.
  133. 2. **Spec.** Read the catalog, test tactics for stage fit against their `requires` and `produces`, draft 5-10 stages. Each stage carries an objective, expected input, focus areas, tactic, numeric completion criteria, backtrack condition, and execution steps. No research runs while this phase is open.
  134. 3. **Execution.** Only after user approval. Each item loads its tactic as a skill rather than doing the work inline, then appends one checkpoint.
  135. ### State: Append-Only Checkpoints
  136. Every tactic and SOP returns the same eight Delta fields:
  137. ```text
  138. findings · evidence_updates · hypothesis_updates · assumption_updates
  139. uncertainties · decisions · open_questions · recommended_jumps
  140. ```
  141. They append to a per-phase context file as numbered checkpoints, and earlier checkpoints are never edited. A plan change is a new `decisions` event, so the reasoning behind the current plan stays readable after the plan itself moves on.
  142. `SpecView` — the live plan — is a projection rebuilt by replaying those events, never persisted as a second object. Plan and history cannot drift apart because there is only one record.
  143. Recovery follows from that: read `context/INDEX.md`, find the phase file, replay to the latest complete checkpoint, resume at the first incomplete item. A `partial` checkpoint is not a resume point — read its open questions, then rebuild from the last complete one. No resume command exists because there is no second progress tracker to resume from.
  144. ### Why Pure Markdown
  145. 1. **Zero infrastructure.** No build, no deploy, no runtime beyond the agent itself.
  146. 2. **Universal composability.** Any skill references any other by name. No import resolution, no version conflicts.
  147. 3. **Readable at every level.** Open a file and you know exactly what the agent will do.
  148. 4. **Instant modification.** Change behavior by editing text. No recompile, no cache.
  149. 5. **Native to the executor.** Following precise written instructions is the one thing a coding agent is unambiguously good at.
  150. ## Decoupled From Retrieval
  151. The library binds to no retrieval tool. Across all 271 files there is not one MCP server name, tool name, API key, or `allowed-tools` declaration. Six nodes say so in their own descriptions: acquisition is `host-selected` or `runtime-selected`, and tool choice is *"left to the host AI."*
  152. Retrieve with whatever tools your agent already has and hand the results in. The library owns everything downstream — what counts as adequate coverage, whether saturation holds, whether a gap is genuine or an artifact of how the evidence was gathered, and when to stop.
  153. Swapping providers therefore needs no skill edit. `trace-citation-neighborhood` chains citations *"independent of the retrieval tool used."*
  154. ## Recommended MCP Servers
  155. A starting point, not a dependency list. DARE requires none of them and reads no config file — install what your own work needs.
  156. | Server | Package | Use for |
  157. | --- | --- | --- |
  158. | **alphaxiv** | — (HTTP, keyless) | arXiv search, paper Q&A, PDF queries, author and affiliation lookup. Academic default |
  159. | **semantic-scholar** | [`@yogsoth-ai/semantic-scholar-mcp`](https://github.com/yogsoth-ai/semantic-scholar-mcp) | Paper lookup, recommendations, and the citation graph as traversable edges — the one server here that exposes it |
  160. | **pubmed** | — | Published biomedical and life-science literature, which arXiv does not cover |
  161. | **biorxiv** | [`biorxiv-mcp`](https://github.com/yogsoth-ai/biorxiv-mcp) | bioRxiv preprint full text — DOI to clean markdown via the official `.meca` TDM archive. Free Europe PMC search; full text reads a Requester-Pays bucket with your own AWS key |
  162. | **medrxiv** | [`medrxiv-mcp`](https://github.com/yogsoth-ai/medrxiv-mcp) | medRxiv preprint full text, same path. With biorxiv it closes the gap between arXiv and what is already published |
  163. | **perplexity** | [`@perplexity-ai/mcp-server`](https://www.npmjs.com/package/@perplexity-ai/mcp-server) | Academic and web both — `perplexity_research` for multi-step literature reconnaissance with `search_domain_filter`, `perplexity_search` and `perplexity_ask` for ranked or synthesized web answers |
  164. | **brave-search** | `@brave/brave-search-mcp-server` | Web, news, and local search plus LLM context. Web default |
  165. | **tavily** | `tavily-mcp` | Web search tuned for LLM consumption |
  166. | **keenable** | — (HTTP, keyless) | Web search plus page fetch returning clean markdown. No install, no key |
  167. | **you** | — (HTTP, keyless) | Web search on the free profile. No install, no key |
  168. | **apify** | `@apify/actors-mcp-server` | Full-page scraping and sources a search API will not reach |
  169. | **wiki-vault** | [`@yogsoth-ai/wiki-vault`](https://github.com/yogsoth-ai/wiki-vault) | Persistent research knowledge graph — BM25 search, typed edges, traversal. Pairs with the STRUCTURING family |
  170. Keenable and You.com need neither an install nor a credential, so an otherwise unconfigured agent still has working web retrieval. If your work routes through `trace-citation-neighborhood`, check that at least one configured server returns citations and references as edges rather than metadata alone.
  171. ## Repository Structure
  172. ```text
  173. de-anthropocentric-research-engine/
  174. ├── skills/ # 271 skills, flat, one directory each
  175. │ ├── dare/ # entry shell
  176. │ ├── research-catalog/ # 51-tactic capability menu
  177. │ ├── write-research-spec/ # spec construction
  178. │ ├── execute-research-spec/ # spec execution + checkpoints
  179. │ └── [267 tactic and SOP nodes] # the graph
  180. ├── .github/ # CI: structural and safety checks
  181. ├── assets/
  182. │ ├── yogsoth-logo.svg
  183. │ ├── the-research-engine.png # figure: control plane, graph, state
  184. │ ├── research-is-not-a-pipeline.png # figure: arsenal vs pipeline
  185. │ ├── contracts-define-completion.png # figure: the five parts of a tactic
  186. │ ├── no-dare-runtime.png # figure: the agent is the runtime
  187. │ └── DE-ANTHROPOCENTRIC.md # the philosophical argument in full
  188. ├── skills.sh.json # skill grouping for the skills.sh listing
  189. ├── package.json # metadata only; no dependencies
  190. ├── README.md
  191. ├── README.zh-CN.md # Simplified Chinese README
  192. ├── SECURITY.md
  193. └── LICENSE
  194. ```
  195. Every skill directory holds exactly one file, `SKILL.md`.
  196. `context/` is a runtime artifact the executing agent creates in your own project, not part of this repository.
  197. ## License
  198. [Apache-2.0](LICENSE)
  199. ---
  200. *The orchestrator of the [Yogsoth AI](https://github.com/yogsoth-ai) research ecosystem. Built by [Pthahnix](https://github.com/Pthahnix).*

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.