org-research
Skill0 starsChanged 36 days agoMITView on GitHub
Verify facts about the organization being applied to, and never invent one. Use during /targets, /whyus, /extract, and any time an application needs a specific detail about the reader's organization.
In this repository9 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- application-typesSkill
- application-writingSkill
- honesty-guardSkill
- interviewing-youSkill
- mock-interviewSkill
- no-slop-writingSkill
- reader-povSkill
- unasked-questionsSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
Every fact you write into an applications/ file carries a source and a date:
- Distinctive: line 34
Never move something from the second block to the first because it is probably true.
≈ “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
Researching the org
The rule against inventing facts about the applicant has a twin that is easier to break and just as expensive: never invent a fact about the organization.
A hallucinated project name, a made-up alum, a client they never had, a professor's research described from the title alone. Every one of these goes into a "why us" answer and gets caught by the one person in the room who was actually there. It is worse than a generic answer, because generic reads as lazy and wrong reads as dishonest.
The standard
Every fact you write into an applications/ file carries a source and a date:
- Runs four pro-bono projects a semester, one per team.
Source: campusconsulting.example.edu/projects, checked 2026-09-02.Anything you cannot source goes in a separate block:
## Unverified, do not use
- Possibly partnered with the county food bank in 2025. Could not confirm. Ask a
member, or drop it.Never move something from the second block to the first because it is probably true.
Where facts actually come from
In descending order of both reliability and value:
- The applicant's own contact with them. An info session, a conversation with a
member, a workshop they attended. Logged in
you/CHATS.md. Unfakeable and nobody else has it. - The org's own current material. Their site, their application form, their handbook, their published rubric or funding priorities, a lab page, a job posting.
- Primary documents they produced. A paper, an annual report, a case study, a product, a public deck.
- First-party posts. Their LinkedIn, Instagram, or newsletter. Check the date.
- Third-party coverage. Useful, and the place errors enter. Attribute it.
Anything you half-remember about an organization from training rather than from a source you just read is not a fact. This is the specific failure mode to watch in yourself, and it is strongest for well-known organizations, where a confident wrong detail is most likely and most damaging.
Things that go stale and get people caught
- Officers, leads, and advisors. People graduate and rotate every year.
- Whether a professor is taking students this cycle.
- Deadlines and cycles carried over from last year's calendar.
- Programs that ended. Naming a discontinued initiative as a reason you are applying is a bad look that a current member spots instantly.
- Funding priorities, which funders revise.
Anything with a person's name or a date on it needs a current source, not a plausible one.
Reading a published rubric
Fellowships, grants, and many student orgs publish what they score. When one exists, it outranks every general instinct about application writing in this repo. Copy the criteria verbatim into the org's file and map each prompt to the criterion it serves.
Most applicants skim the criteria once and never look again. Mapping answers to them is close to free and almost nobody does it.
When research comes up empty
Say so plainly rather than padding the file. Then name the two fastest fixes: attend
the thing on the date in DEADLINES.md, or message one member and run /chat. Thirty
minutes of either produces better material than any amount of searching, and it is the
only way to get a first-tier specific.
An honest "I could not verify anything specific about this org" is a useful finding. It usually means the "why us" answer has to be built from what the applicant will do for them rather than from what they already are.
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.

