building-report
Skill0 starsChanged 41 days agoMITView on GitHub
Produce a complete, sourced research note on one Dubai apartment building — prices, rents, gross yields, growth and evidence for every unit size — using the Dubai Wealth Index tools. Use when the user asks to "tell me about", assess, research, write up or do due diligence on a named Dubai building or tower, or wants a report, spreadsheet or chart about one.
In this repository8 other files it gives its agents.
- check-a-claimSkill
- check-asking-priceSkill
- cite-dubai-property-figuresSkill
- get-startedSkill
- request-infoSkill
- shortlist-and-compareSkill
- short-term-rentalSkill
- track-saved-buildingsSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 2 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Call search with the name the user gave.
- Very specific: line 13
Check resolution, score and locality before accepting a hit.
- Very specific: line 17
If two hits are close, or get_building returns resolved.building.confidence of fuzzy, ask the user which building they mean before going …
- Very specific: line 22
Call get_building with detail: "report".
- Very specific: line 35
If a headline rests on few sales (a count under about 10), or the user asks how solid a figure is, call get_building_transactions for the …
- Distinctive: line 39
Keep the same bedroom type, cohort and window.
- Distinctive: line 54
Never add a figure the tools did not return, and never describe a gross yield as a net return.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Report on one Dubai building
Explicit user instructions take priority over this workflow.
1. Confirm the building
- Call
searchwith the name the user gave. - Check
resolution, score and locality before accepting a hit. AmatchedNamealone only names a matched alias; fuzzy candidates need confirmation. For an exact alias match, say so in the answer: "Vera Residences is registered as Vera Tower, Business Bay." - If two hits are close, or
get_buildingreturnsresolved.building.confidenceoffuzzy, ask the user which building they mean before going further.
2. Read the whole building once
Call get_building with detail: "report". Do not call it again per bedroom
type — the report already covers every publishable one. It returns:
identity— registered name, area, DLD project. Quote the registered name.bedrooms[]— for each publishable unit size: sale and rent medians, gross yield and its price basis, short-term and off-plan figures where published,growthwith both comparison windows, the area benchmark and a 6-year trend.growthWindows— the dates and bases every growth figure compares.suppressed[]— each figure that did not clear its evidence gate, and why.limitations[]— the caveats that travel with the figures.
3. Check the evidence when it matters
If a headline rests on few sales (a count under about 10), or the user asks
how solid a figure is, call get_building_transactions for the same building
and say whether the median rests on many comparable deals or a few scattered
ones. Check factsEdition against metricsEdition before reconciling counts.
Keep the same bedroom type, cohort and window. Label pooled rents explicitly;
an empty building-level rent list does not mean the rent pool has no contracts.
4. Write it up
- Lead with the registered name, area and data date (
dataAsOf). - For each unit size: price, rent, gross yield, and the sample behind each.
- Growth: give the percentage with both windows, e.g. "+8.2% (median AED 2,040 /sqft on 46 sales vs AED 1,885 on 51 in the prior 12 months)".
- List the suppressed figures as "not published:
<reason>", never as zero. - Close with the limitations that apply, and link
attribution/ thecanonicalUrl. - For a chart or spreadsheet, use the report's
trendrows and per-bedroom figures as the data; do not interpolate missing years.
Never add a figure the tools did not return, and never describe a gross yield as a net return.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

