business-agent-skills
AGENTS.md1 starsChanged 2 months agoMITView on GitHub
Seven Agent Skills for business, career and engineering work, built on the Agent Skills open standard. Skills live at skills/<category>/<skill-name>/SKILL.md. Categories organise the repo only — installing flattens them into one directory, so skill names must be unique across every category. A skill is one directory. Which file a thing belongs in follows from when it is read: SKILL.md is loaded in full on every run, so its length is a cost paid every time. Every skill has a word…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository8 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- company-researchSkill
- career-directionSkill
- interview-prepSkill
- pay-checkSkill
- role-fitSkill
- orchestrateSkill
- learning-pathSkill
The file
We split this AGENTS.md into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
No two sections in one skill may be read as the same step.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
Agent instructions
Seven Agent Skills for business, career and engineering work, built on the Agent Skills open standard.
Skills live at skills/<category>/<skill-name>/SKILL.md. Categories organise the repo only —
installing flattens them into one directory, so skill names must be unique across every category.
What goes where
A skill is one directory. Which file a thing belongs in follows from when it is read:
| Path | Holds | Read |
SKILL.md | what the agent must choose — the steps, the questions, when to stop | in full, every time the skill fires |
references/<topic>.md | what the agent must construct — the shape of an output, how a bundled file works | only at the step that needs it |
scripts/ | what can be checked rather than read | run, never read |
assets/ | templates and fixtures the skill ships | as used |
evals/ | cases that test the skill's own script or page | never by the skill; left out of the package |
SKILL.md is loaded in full on every run, so its length is a cost paid every time. Every skill
has a word budget: 1,000 by default, or whatever budget: in its frontmatter declares.
build_skills.py counts the file and fails the build when it is over. An addition that would
breach the budget means a construct-shaped section moves to references/ first — not that the
addition is dropped. Raise the budget only when nothing is left to move, and say why in the
pull request.
A rule a script can enforce belongs in the script, with one line in SKILL.md to run it. Prose
restating a shipped check is one rule in two places.
No two sections in one skill may be read as the same step. An agent told to add something to the step that asks the user for input must have exactly one candidate section.
Why this shape, and what was rejected: the decision records.
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.