Research brief — instructions for SELIS V2 research chats
Read CLAUDE.md first. It describes SELIS, the rules of
this project and the order of authority. This file gives the rules
shared by all research topics. Each topic has its own file in
research/topics/.
How to use
Start one chat per topic with: "Follow research-brief.md and
research/topics/
The topic list is not fixed. The topics are whatever files exist in
research/topics/; new ones are added as findings come in
(see "Adding a topic"). The first set is:
| Topic | File | Run |
|---|---|---|
| A. Competitors and prior ESA-funded work | research/topics/A-competitors.md |
First round, in parallel |
| B. Data sets | research/topics/B-data-sets.md |
First round, in parallel |
| C. Users and their needs | research/topics/C-users-and-needs.md |
First round, in parallel |
| D. Available technology | research/topics/D-technology.md |
First round, in parallel |
| E. Feature opportunities | research/topics/E-features.md |
After A to D |
Purpose
Research supplies evidence for the PECS proposal, and it comes before the scope is fixed. The findings will shape what the Type A activity is. Do not assume a scope: report what the evidence supports and what it rules out.
Every finding must help a specific proposal section or a scope decision. If something is interesting but would change neither, leave it out.
Output: one file per topic
Save as
research/findings/<same name as the topic file>, at
most two pages, in this order. A topic file may add or replace output
items; where it does, the topic file wins.
- Feeds: the proposal section(s) this supports, by template number (given in the topic file).
- Key findings: 5 to 10 points, each with its source link. Most important first.
- Evidence table: the comparable facts for the topic (columns are given in the topic file).
- What this means for the Type A scope: 3 to 5 points on what the evidence supports, what it rules out, and how SELIS should be positioned.
- Ready-to-paste sentences: five sentences in proposal style (specific, neutral, British English), each traceable to a cited source.
- Open risks and unknowns: what could not be confirmed, what contradicts the current SELIS description, what needs a decision from Ivaylo.
- Suggested follow-up topics: at most three questions that this research raised and could not answer within its own topic, each with one line on why it matters for the scope or for a proposal section. Write "none" if there are none. Do not research them in this chat.
- Sources: full list with links and access date.
Rules
- Scope: the Type A activity focuses on the lunar South Pole
(decided). Nothing else about the scope is decided. The pipeline, data
sets and tools described in
CLAUDE.mdare a baseline hypothesis; challenge them where evidence supports a better option. - Type A ends at TRL 3: a proof of concept of critical functions, at most EUR 200,000 and 9 to 15 months. Judge feasibility against that, not against the long-term vision.
- Cite only sources that were actually opened and checked in this chat. No citations from recall.
- Prefer primary sources: ESA and agency pages, mission archives and data-set documentation, peer-reviewed papers, company documentation. Use news and blogs only when nothing better exists, and say so.
- Give dates. Anything about who is active, what is funded or what is available must state as of when.
- Separate fact from judgement. Mark estimates as estimates.
- If something cannot be found, write "not found" and where you looked. Never fill a gap with a plausible guess.
- If a finding contradicts
CLAUDE.mdor the RFI draft, put it at the top of section 6. - Do not write proposal sections, redesign the product or widen the topic.
- Read the existing files in
research/findings/before starting, so the work builds on them and does not repeat them. If a finding there conflicts with yours, say so and cite both.
Adding a topic
Ivaylo decides which follow-up topics to open. A new topic is one new
file in research/topics/, named with the next free letter
and a short name, for example F-site-selection.md. Add a
row for it to the table in "How to use".
A topic file contains, in this order:
- Title: "Topic
— ", then the line: Follow research-brief.mdfor the output format and rules. Save the result asresearch/findings/<file name>. - Origin: the finding or decision that raised it, with the file it came from. Omit for the first-round topics.
- Depends on: the findings to read first, if any.
- Feeds: the proposal sections, by template number, or the scope decision it informs.
- Find: what to look for, as specifically as possible.
- Table columns: the columns of the evidence table.
- Answer: the one question the research must answer.
Keep a topic narrow enough to answer in one chat and two pages. If a topic needs more, split it.
A topic can also be a deeper second pass on an existing one. Name it
after the original, for example
B2-south-pole-stereo-coverage.md, and list the original
finding under "Depends on".