| name | revela-research |
| description | Research from an existing or emerging Revela spec.md and public/workspace sources, using active domain/design guidance to save source-linked findings and hand off deck-plan.md. |
Revela Research
Use this skill when the user asks to start from a goal, inspect local materials, research missing inputs, gather public support, save findings, find source-linked examples/assets, or prepare the deck planning handoff.
Contract
- Research is the source-preparation workflow for Codex Revela.
- Prefer root-level
spec.md as the demand contract. If it is missing or the objective is unclear, route to revela-spec unless the user gave enough intent to research immediately.
- Research output is saved under
researches/**/*.md and, when the user goal is a deck and materials are sufficient, handed off as deck-plan.md.
- Local materials are only usable after direct text review or extracted read-view review.
- Active/requested domain guidance informs audience, decision framing, claim standards, evidence expectations, objection/risk interpretation, and research-gap priority.
- Domain guidance is not evidence and must never be cited as proof for factual claims.
- Active/requested design tools define page-template vocabulary, template slots, design rules, and deck-plan styling context.
deck-plan.md is the formal research-to-make-deck handoff when a deck objective is sufficiently supported.
- Do not create deck artifacts, a Narrative Vault, or canonical evidence bindings during research.
- Do not invent URLs, quotes, page references, numbers, caveats, or licenses.
Preconditions
- The user provides at least one of: existing
spec.md, objective, topic, audience, decision/action, source materials, or deck intent.
- If intent is unclear, inspect the workspace first and ask only the smallest missing high-impact questions.
Inputs
- Root-level
spec.md when present.
- User objective, constraints, audience, decision/action, and language preference when available.
- Workspace materials, extracted material read views, existing
researches/**/*.md, existing deck-plan.md, and assets/.
- Active or user-requested domain guidance.
- Optional external public sources when needed and allowed.
Required Tool Order
- Call
revela_domain_list.
- Call
revela_domain_read for the active domain or user-requested domain.
- Read
spec.md when present and use it to scope material review, findings, and deck-plan handoff.
- Call
revela_prepare_local_materials.
- For Office/PDF sources, read the provided
allowedReadPath / read_view_path; if missing, call revela_extract_document_materials.
- Read original text/Markdown/CSV files or extracted read views before treating a material as usable.
- Call
revela_record_material_review for each reviewed Office/PDF source.
- Call
revela_check_material_intake before reporting research readiness.
- Use external research only for public facts, user-authorized questions, or gaps not covered by local materials or
spec.md.
- Save useful findings with
revela_research_save.
- For deck goals with sufficient materials, run Planning Handoff:
- Call
revela_design_list.
- Call
revela_design_read with section: "rules" for the active/requested design.
- Call
revela_design_inventory and use its pageTemplates summary as the public design vocabulary.
- Call
revela_list_page_templates before choosing template ids.
- Write
deck-plan.md directly from reviewed materials, saved findings, assets, user intent, active domain framing, and built-in page templates.
- Call
revela_read_deck_plan after writing deck-plan.md.
- If diagnostics report
sourceLinks or template id issues, patch deck-plan.md directly and call revela_read_deck_plan again.
Finding Requirements
Saved research should use stable, reusable Markdown blocks. Evidence findings use:
## Finding: <stable-id>
Source: <source name and date when known>
URL: <source URL when available>
Location: <page/slide/sheet/section when known>
Quote/Snippet: <short exact quote or compact snippet; note when no exact quote is available>
Supports: <narrow support scope or intended slide/source context>
Evidence boundary: <internal guardrail; what this finding does not prove>
Strength: <strong|directional|weak|context-only>
Deck use: <where this belongs in deck planning>
Display note: <optional short user-facing scope note for captions/source notes>
Use synthesis blocks to turn multiple findings into decision-relevant interpretation before deck planning:
## Synthesis: <stable-id>
Question answered: <research question this synthesis resolves>
Basis: <finding ids, source files, or URLs used>
Interpretation: <what the evidence means when read together>
So what: <why this matters for the audience or decision>
Decision implication: <what should change in the recommendation, story, or slide argument>
Confidence: <high|medium|low>
Alternative reading: <plausible competing interpretation or contradiction>
Evidence boundary: <internal guardrail; what this synthesis must not overclaim>
Deck use: <where this belongs in deck planning>
Display note: <optional short user-facing scope note>
Use ## Analysis: <stable-id> for user/LLM analytical frameworks, ## Implementation Note: <stable-id> for render/data/API contracts, ## Asset Lead: <stable-id> for image/logo/media leads, and ## Gaps for missing or insufficient source support.
Each saved evidence finding should include:
- Source URL or workspace path.
- Quote/snippet or explicit note when no exact quote is available.
- What it supports.
Evidence boundary for internal support limits, unsupported scope, or uncertainty.
Deck use for likely planning placement.
- Optional
Display note for short audience-facing scope text.
- Date checked.
- Optional image/logo/screenshot leads with known source and license/attribution status.
If a finding is context only, label it as context and do not present it as proof. Internal boundaries must not be mechanically copied into deck text; use Display note for default visible caption/source-note scope, and expose Evidence boundary only when needed to avoid a misleading audience conclusion.
Do not use raw findings as the default deck argument. For deck goals, synthesize findings first; findings provide evidence basis, while Synthesis provides the interpretation, decision implication, and audience takeaway that should drive deck-plan.md.
Outputs
researches/{topic}/{filename}.md
- Material review records for reviewed Office/PDF sources.
deck-plan.md when the user goal is a deck and reviewed materials/findings are sufficient for a traceable plan.
- Source limitations and unresolved gaps.
- A clear statement of whether
revela-make-deck can proceed or whether more research is needed.
Planning Handoff
Use this final stage only for deck goals. If sources are too thin, report unresolved inputs and source limitations instead of drafting unsupported slides.
Every deck-plan.md handoff should include Cover, Table of Contents, Closing, 3-5 chapter headings, explicit slide ranges, and --- slide separators under ## Slides.
Each non-structural slide block must include:
- Slide title and role when relevant.
#### Content Plan
- In
#### Content Plan: Claim, Reasoning, Audience takeaway, Evidence basis, and Boundary handling.
#### Source Links for materials, finding-level references when available, assets, URLs, and caveats.
- A
template field in slide frontmatter using a built-in page template id.
- Optional
#### Template Content JSON when a template needs structured seed content.
- Base slide arguments on
Synthesis blocks when available; use finding text as evidence/source context, not as default body copy.
- Use
Display note for short visible caption/source-note scope.
- Keep
Evidence boundary internal unless it is required to avoid a misleading audience conclusion.
Analysis and Implementation Note entries may support deck structure or rendering, but must not be cited as external factual proof.
- Unresolved inputs, source limitations, and user review notes instead of AI-authored caveat/risk judgement.
Do not duplicate the same child as both nested and top-level. Do not add source links that were not reviewed or saved during research.
Must Not
- Do not generate
revela-narrative/.
- Do not write
spec.md; route demand changes to revela-spec.
- Do not write
decks/*.html.
- Do not bind findings into a Narrative Vault or canonical evidence graph.
- Do not treat domain guidance as source evidence.