panjose/Co-Scientist / skills/research-config
research-config
VerifiedGenerate a `ResearchPlan` from `input.md` or equivalent raw research input.
research-config
Goal:
- Generate a
ResearchPlanfrominput.mdor equivalent raw research input.
Inputs:
input.mdor equivalent raw research brief text- optional
state/START_REQUEST.json - optional
RUN_POLICY.yaml
Outputs:
research_plan/RESEARCH_PLAN.jsonresearch_plan/RESEARCH_PLAN.md
Context Loading:
- Read
input.mdfirst. Treat it as the canonical user brief for the run. - If
state/START_REQUEST.jsonexists, use it only as auxiliary context for how the run was started. Do not let it override the actual brief text ininput.md. - If
RUN_POLICY.yamlexists, use it only as supporting context for expected run style or emphasis. Do not copy policy values into the research goal unless the brief implies them. - Extract three things from the research input:
- the primary
research_goal preferencesthat define what a strong hypothesis should optimize forconstraintsthat all downstream hypotheses must satisfy
- the primary
Execution Prompt Contract:
- System Intent:
- You are the run's research-plan structuring layer.
- Your job is to convert raw user research input into a concise, stable
ResearchPlanContract.
- Required Reasoning Focus:
- Extract the main scientific objective faithfully when it is explicit.
- If the goal is underspecified, synthesize the shortest accurate formulation that preserves the user's intent.
- Derive
preferencesas evaluation criteria for hypothesis quality. - Derive
constraintsas hard boundaries for downstream generation and review. - If
preferencesorconstraintsare not explicit, infer only reasonable defaults from the research domain and stated goal.
- Do Not Do:
- Do not invent specific scientific facts that are not present or reasonably implied.
- Do not turn broad domain assumptions into narrow claims unless the brief clearly supports them.
- Do not emit verbose analysis or chain-of-thought style discussion.
- Do not exceed five preferences or five constraints.
- Output Shape:
- Produce a canonical
ResearchPlanContract. research_goalmust be a single natural-language string, ideally no more than three sentences.- Each
preferencesandconstraintsitem must be short, specific, and directly useful to downstream skills.
- Produce a canonical
Execution Steps:
- Open
skills/shared-references/schema-index.md, then readpackages/agent_contracts/research_plan.pybefore writingresearch_plan/RESEARCH_PLAN.json. - Read
input.md. - If present, read
state/START_REQUEST.jsonandRUN_POLICY.yamlas contextual hints only. - Identify the main research objective.
- Extract or infer up to five
preferences. - Extract or infer up to five
constraints. - Write the canonical
research_plan/RESEARCH_PLAN.json. - Write the companion
research_plan/RESEARCH_PLAN.md. - Run validation before declaring the skill complete.
Artifact Rules:
research_plan/RESEARCH_PLAN.jsonmust validate against the canonical sharedResearchPlanContract.research_plan/RESEARCH_PLAN.mdmust remain a human-readable rendering of the same plan, not a divergent summary.- The JSON artifact is authoritative; the Markdown file is a companion view.
Completion Rule:
- This skill is complete only when
research_plan/RESEARCH_PLAN.jsonandresearch_plan/RESEARCH_PLAN.mdboth exist and the JSON artifact is valid for downstream consumption.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |