skills
AGENTS.md6 starsChanged 2 months agoMITView on GitHub
Guidance for agents working in this repository. Humans: read CONTRIBUTING.md instead; it is the same rules with the reasoning. Markdown skills for agents that run paid research on Glasser. It is also a Claude Code plugin marketplace (.claude-plugin/) and is installable with npx skills add glasser-ai/skills. There is no build step and no runtime code outside .github/scripts/. The Glasser mechanism (auth, search, inspect, run, balance rules) is documented at https://glasser.ai/SKILL.md and shipped by glasser-ai/plugins. Nothing here duplicates it. Node 20…
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 31,700 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 repository20 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ad-intelligenceSkill
- ai-search-visibilitySkill
- competitor-researchSkill
- ecommerce-ad-creative-researchSkill
- ecommerce-amazon-opportunity-scanSkill
- ecommerce-competitor-price-monitorSkill
- ecommerce-review-insightsSkill
- glasserSkill
- influencer-prospectingSkill
- investor-diligenceSkill
- product-demand-researchSkill
- prospect-listSkill
- seo-auditSkill
- seo-find-new-keywordsSkill
- seo-free-tool-opportunitiesSkill
- seo-new-keyword-competitionSkill
- social-trendsSkill
- startup-analysisSkill
- youtube-kol-finderSkill
The file
We split this AGENTS.md into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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 30
Every SKILL.md carries the block in templates/glasser-prereq.md verbatim, between the H1 and the first ##.
- Distinctive: line 32
No HTML comments, no CJK characters, under 500 lines.
- Distinctive: line 33
No provider names in skill names or process steps.
- Distinctive: line 34
No execution channels: a skill ends at a result, it does not send, post, publish, or buy.
≈ “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
AGENTS.md
Guidance for agents working in this repository. Humans: read CONTRIBUTING.md instead; it is the same rules with the reasoning.
What this repository is
Markdown skills for agents that run paid research on Glasser. It is also a Claude Code plugin marketplace (.claude-plugin/) and is installable with npx skills add glasser-ai/skills. There is no build step and no runtime code outside .github/scripts/.
The Glasser mechanism (auth, search, inspect, run, balance rules) is documented at https://glasser.ai/SKILL.md and shipped by glasser-ai/plugins. Nothing here duplicates it.
Layout
.claude-plugin/
plugin.json plugin manifest; "skills": "./skills"
marketplace.json marketplace manifest; one plugin, source "./"
.github/
scripts/validate.mjs check every skill and the repo-level invariants
scripts/sync.mjs regenerate the README table, marketplace count and per-skill prerequisite block
workflows/ validate on PR and push to main; sync check on main
skills/<name>/SKILL.md one skill per directory
templates/ SKILL.template.md and glasser-prereq.md; outside skills/ on purposeInvariants
namein frontmatter equals the directory name.- Frontmatter has exactly
name,description,metadata.version,metadata.category. descriptionis on one line withdescription:, 1 to 1024 characters.- Every
SKILL.mdcarries the block intemplates/glasser-prereq.mdverbatim, between the H1 and the first##. - No
](../or](/links; skills are copied out alone. - No HTML comments, no CJK characters, under 500 lines.
- No provider names in skill names or process steps. Describe the capability to search for and what to confirm on inspect.
- No execution channels: a skill ends at a result, it does not send, post, publish, or buy.
- No
.mcp.jsonhere, notools/directory, noshared/directory. skills/glasser/is the one mirror of glasser.ai/SKILL.md: body byte-identical to the source, frontmattername,description,version,metadata.source,metadata.category, no prerequisite block. Regenerate it withnode scripts/sync-glasser-skill.mjs; never edit it by hand.- Nothing under
templates/is namedSKILL.md; installers would treat it as a skill. - README table, marketplace skill count and every skill's prerequisite block are generated;
sync.mjs --checkenforces all three.
Commands
node .github/scripts/validate.mjs # all checks; exit 1 on any error
node .github/scripts/sync.mjs # rewrite README table, marketplace count, prerequisite blocks
node .github/scripts/sync.mjs --checkNode 20 or later, no dependencies.
Adding a skill
cp templates/SKILL.template.md skills/<name>/SKILL.mdand write it.node .github/scripts/sync.mjsnode .github/scripts/validate.mjs- Branch
skills/<name>, conventional commitfeat(<name>): add skill, PR with the new-skill template.
Changing a skill
Bump metadata.version in the same PR. Patch for fixes, minor for new dimensions or triggers. To change the prerequisite wording, edit templates/glasser-prereq.md and run sync.mjs; commit the source and the rewritten skills together.
Versioning
plugin.json version and marketplace.json metadata.version share one number and move together: minor for a new skill, patch for changes to existing ones, major for repo-wide restructuring. Per-skill versions live in each SKILL.md and move independently.
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.

