registry.yaml

version: 0.9.0
backends:
- name: dartmouth
  kind: openai_compatible
  auth_env_vars:
  - DARTMOUTH_CHAT_API_KEY
  daily_quota_estimate: 100
  is_paid: false
- name: local
  kind: local_transformers
  auth_env_vars: []
  is_paid: false
agents:
- name: brainstorm
  purpose: Generate one short raw idea seed for a research project in a given field.
  inputs:
  - idea
  outputs:
  - idea
  prompt_path: agents/prompts/brainstorm.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: flesh_out
  purpose: Expand a raw idea seed into a structured idea grounded in primary sources.
  inputs:
  - idea
  outputs:
  - idea
  prompt_path: agents/prompts/flesh_out.md
  prompt_version: 1.2.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools:
  - lit_search
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_question_validator
  purpose: Audit a fleshed-out idea for research-question quality (phenomenon-vs-method,
    circularity, triviality, narrowing) before idea_selector promotion.
  inputs:
  - idea
  outputs:
  - idea
  - project_state
  prompt_path: agents/prompts/research_question_validator.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: idea_selector
  purpose: Decide whether a fleshed-out idea should be promoted to a Spec Kit project.
  inputs:
  - idea
  outputs:
  - project_state
  prompt_path: agents/prompts/idea_selector.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: project_initializer
  purpose: Bootstrap a per-project Spec Kit scaffold and render a project constitution.
  inputs:
  - idea
  outputs:
  - project_state
  prompt_path: agents/prompts/project_initializer.md
  prompt_version: 1.2.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: librarian
  purpose: Canonical literature-search + verification agent (spec 005/006). 3 backends
    (Semantic Scholar, arXiv, TheoremSearch) + an LLM math-classifier for non-math
    fields. Tool-style.
  inputs:
  - idea
  outputs:
  - idea
  prompt_path: agents/prompts/librarian.md
  prompt_version: 1.6.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 1800
  paid_opt_in: false
- name: specifier
  purpose: Drive /speckit.specify for the project; draft spec.md from the idea.
  inputs:
  - idea
  outputs:
  - technical_design
  prompt_path: agents/prompts/specifier.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: clarifier
  purpose: Drive /speckit.clarify; resolve [NEEDS CLARIFICATION] markers in spec.md.
  inputs:
  - technical_design
  outputs:
  - technical_design
  prompt_path: agents/prompts/clarifier.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: planner
  purpose: Drive /speckit.plan; produce plan.md, research.md, data-model.md, quickstart.md,
    contracts/.
  inputs:
  - technical_design
  outputs:
  - implementation_plan
  prompt_path: agents/prompts/planner.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: code_adapter
  purpose: Turn a vendored GPU-heavy ingested-paper repo into a simplified, CPU-runnable
    adaptation under code/ + quickstart.md so the execution gate can run it.
  inputs:
  - code
  outputs:
  - code
  prompt_path: agents/prompts/code_adapter.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: tasker
  purpose: Drive /speckit.tasks then /speckit.analyze; resolve every analyze finding.
  inputs:
  - implementation_plan
  outputs:
  - implementation_plan
  prompt_path: agents/prompts/tasker.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: implementer
  purpose: Drive /speckit.implement; complete the next incomplete task within the
    leaf budget.
  inputs:
  - implementation_plan
  outputs:
  - code
  - data
  prompt_path: agents/prompts/implementer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: research_reviewer
  purpose: Read implemented research artifacts and emit a structured review record
    with verdict accept/minor_revision/full_revision/reject.
  inputs:
  - code
  - data
  - implementation_plan
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: reference_validator
  purpose: Verify every cited URL/arXiv/DOI/dataset against its primary source; gate
    accept transitions on verification.
  inputs:
  - technical_design
  - implementation_plan
  - paper
  - code
  outputs:
  - review
  prompt_path: agents/prompts/reference_validator.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools:
  - citation_fetcher
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: paper_initializer
  purpose: Bootstrap a per-project paper Spec Kit scaffold and render a paper-stage
    constitution.
  inputs:
  - implementation_plan
  - code
  - data
  outputs:
  - project_state
  prompt_path: agents/prompts/paper_initializer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: paper_specifier
  purpose: 'Drive /speckit.specify for the paper artifact: sections, required figures,
    claim list, target venue style.'
  inputs:
  - implementation_plan
  - code
  - data
  outputs:
  - paper
  prompt_path: agents/prompts/paper_specifier.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_clarifier
  purpose: Drive /speckit.clarify on the paper spec; resolve [NEEDS CLARIFICATION]
    markers.
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_clarifier.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_planner
  purpose: Drive /speckit.plan for the paper; produce plan.md, research.md, data-model.md,
    quickstart.md, contracts/.
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_planner.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: paper_tasker
  purpose: Drive /speckit.tasks + /speckit.analyze for the paper; emit tasks with
    [kind:...] tokens.
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_tasker.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: paper_implementer
  purpose: Drive /speckit.implement for the paper; dispatch each task to the appropriate
    sub-agent by [kind:...] token.
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_implementer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: paper_writing
  purpose: Compose LaTeX prose for one paper section/subsection at a time.
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_writing.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools:
  - lit_search
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: paper_figure_generation
  purpose: Generate Python plotting code that produces a figure from real data; runs
    in the code sandbox.
  inputs:
  - data
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_figure_generation.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools:
  - code_sandbox
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: paper_statistics
  purpose: Compute and interpret one inferential analysis per task; writes both the
    analysis script and LaTeX prose.
  inputs:
  - data
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/paper_statistics.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools:
  - code_sandbox
  wall_clock_budget_seconds: 900
  paid_opt_in: false
- name: proofreader
  purpose: Flag repeated sections, internal inconsistencies, jargon overuse, and logical
    issues in the paper LaTeX source.
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/proofreader.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: latex_build
  purpose: "Compile LaTeX \u2192 PDF; emit a structured failure summary on compile\
    \ errors."
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/latex_build.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: latex_fix
  purpose: Repair LaTeX compilation errors flagged by latex_build; conservative patches
    only.
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/latex_fix.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer
  purpose: Read the assembled paper and emit a review record with verdict accept/minor_revision/major_revision_writing/major_revision_science/fundamental_flaws.
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: status_reporter
  purpose: Emit per-run summary, post issue comments, regenerate web/data/projects.json,
    compute SC-001/SC-007/SC-012 metrics.
  inputs:
  - review
  outputs:
  - status_comment
  - project_state
  prompt_path: agents/prompts/status_reporter.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: repository_hygiene
  purpose: Read-only hygiene checker. Verifies gitignore patterns, flags scratch artifacts,
    asserts SC-007 line-count delta.
  inputs:
  - review
  outputs:
  - review
  prompt_path: agents/prompts/repository_hygiene.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: task_atomizer
  purpose: Decompose an over-budget parent task into sub-tasks each fitting the leaf
    wall-clock budget; supports recursive atomization up to MAX_ATOMIZATION_DEPTH.
  inputs:
  - implementation_plan
  - paper
  outputs:
  - implementation_plan
  - paper
  prompt_path: agents/prompts/task_atomizer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: task_joiner
  purpose: Auto-spawned when every sibling sub-task succeeds; merges sub-task outputs
    into the parent task's expected artifact format.
  inputs:
  - implementation_plan
  - paper
  outputs:
  - implementation_plan
  - paper
  prompt_path: agents/prompts/task_joiner.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: paper_reviewer_writing_quality
  purpose: Paper-stage specialist reviewer focusing exclusively on writing quality
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_writing_quality.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_logical_consistency
  purpose: Paper-stage specialist reviewer focusing exclusively on logical consistency
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_logical_consistency.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_claim_accuracy
  purpose: Paper-stage specialist reviewer focusing exclusively on claim accuracy
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_claim_accuracy.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_overreach
  purpose: Paper-stage specialist reviewer focusing exclusively on overreach aspects
    of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_overreach.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_safety_ethics
  purpose: Paper-stage specialist reviewer focusing exclusively on safety ethics aspects
    of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_safety_ethics.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_scientific_evidence
  purpose: Paper-stage specialist reviewer focusing exclusively on scientific evidence
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_scientific_evidence.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_statistical_analysis
  purpose: Paper-stage specialist reviewer focusing exclusively on statistical analysis
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_statistical_analysis.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_code_quality_paper
  purpose: Paper-stage specialist reviewer focusing exclusively on code quality paper
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_code_quality_paper.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_data_quality_paper
  purpose: Paper-stage specialist reviewer focusing exclusively on data quality paper
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_data_quality_paper.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_text_formatting
  purpose: Paper-stage specialist reviewer focusing exclusively on text formatting
    aspects of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_text_formatting.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_figure_critic
  purpose: Paper-stage specialist reviewer focusing exclusively on figure critic aspects
    of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_figure_critic.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: paper_reviewer_jargon_police
  purpose: Paper-stage specialist reviewer focusing exclusively on jargon police aspects
    of the manuscript
  inputs:
  - paper
  outputs:
  - review
  prompt_path: agents/prompts/paper_reviewer_jargon_police.md
  prompt_version: 1.1.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_idea_quality
  purpose: Research-stage specialist reviewer focusing exclusively on idea quality
    aspects of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_idea_quality.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_creativity
  purpose: Research-stage specialist reviewer focusing exclusively on creativity aspects
    of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_creativity.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_implementation_correctness
  purpose: Research-stage specialist reviewer focusing exclusively on implementation
    correctness aspects of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_implementation_correctness.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_implementation_completeness
  purpose: Research-stage specialist reviewer focusing exclusively on implementation
    completeness aspects of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_implementation_completeness.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_code_quality_research
  purpose: Research-stage specialist reviewer focusing exclusively on code quality
    research aspects of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_code_quality_research.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_data_quality_research
  purpose: Research-stage specialist reviewer focusing exclusively on data quality
    research aspects of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_data_quality_research.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: research_reviewer_filesystem_hygiene
  purpose: Research-stage specialist reviewer focusing exclusively on filesystem hygiene
    aspects of the implementation
  inputs:
  - code
  outputs:
  - review
  prompt_path: agents/prompts/research_reviewer_filesystem_hygiene.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: submission_intake
  purpose: "Triage human-submission GitHub issues from the website \u2014 route feedback\
    \ to the right pipeline step, create a project, file a submitted paper, or acknowledge\
    \ \u2014 then comment and close."
  inputs:
  - review
  outputs:
  - status_comment
  - project_state
  prompt_path: agents/prompts/submission_intake.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 300
  paid_opt_in: false
- name: personality
  purpose: "Simulated public-figure personas (spec 008) \u2014 one per 30-min tick,\
    \ rotating over agents/prompts/personalities/. Action \u2208 comment / contribute\
    \ / propose_arxiv / abstain, tagged \"<Name> (simulated)\"."
  inputs:
  - review
  outputs:
  - review
  - status_comment
  prompt_path: agents/prompts/personality.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false
- name: llmxive_implementer
  purpose: "Spec 013: LLM-driven implementer (display: llmXive-implementer-v1.0). Applies revision-spec tasks to paper/source/main.tex with per-task compile gate and rollback, then routes to paper_review."
  inputs:
  - paper
  - implementation_plan
  outputs:
  - paper
  - review
  prompt_path: agents/prompts/implementer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends:
  - local
  default_model: qwen.qwen3.5-122b
  tools: []
  wall_clock_budget_seconds: 1800
  paid_opt_in: false
- name: paper_publisher
  purpose: "Spec 013: deterministic (no-LLM) publisher. Pre-reserves Zenodo DOI, recompiles PDF with final byline + appendix, uploads + publishes, writes publication.yaml, transitions paper_accepted to posted."
  inputs:
  - paper
  outputs:
  - paper
  prompt_path: agents/prompts/implementer.md
  prompt_version: 1.0.0
  default_backend: dartmouth
  fallback_backends: []
  default_model: deterministic-no-llm
  tools: []
  wall_clock_budget_seconds: 600
  paid_opt_in: false