flins
AGENTS.md39 starsChanged 10 months agoMITView on GitHub
Use Bun exclusively. Do NOT use npm, pnpm, or yarn under any circumstances. When operating in Plan Mode: Unresolved Questions: None. Failure to follow this structure is not allowed. All generated code must follow these rules: Strict consistency is required across: Before writing any code: When adding new code: All rules below are NON-NEGOTIABLE. If any applicable skill is not used, the response is INVALID. Planning & Scope - Skill(avoid-feature-creep) Required for feature planning, scope discussion, MVPs, roadmaps, and backlogs.
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 31,439 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
- avoid-feature-creepSkill
- clack-guidelinesSkill
- commander-guidelinesSkill
- Convex AgentsSkill
- Convex Best PracticesSkill
- Convex Component AuthoringSkill
- Convex Cron JobsSkill
- Convex File StorageSkill
- Convex FunctionsSkill
- Convex HTTP ActionsSkill
- Convex MigrationsSkill
- Convex RealtimeSkill
- Convex Schema ValidatorSkill
- Convex Security AuditSkill
- Convex Security CheckSkill
- copy-editingSkill
- copywritingSkill
- programmatic-seoSkill
- seo-auditSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
18 rules. About 0 are common, 18 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈18
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 2
Do NOT use npm, pnpm, or yarn under any circumstances.
- Distinctive: line 6
When operating in Plan Mode:
- Distinctive: line 10
Do not include explanations, prose, or filler.
- Distinctive: line 12
Every plan MUST end with a section titled “Unresolved Questions”.
- Distinctive: line 14
If there are no blockers, explicitly write:
- Distinctive: line 22
All generated code must follow these rules:
- Distinctive: line 25
Do not use comments.
- Distinctive: line 35
Before writing any code:
- Distinctive: line 40
When adding new code:
- Distinctive: line 42
Place it only in logically correct locations.
- Distinctive: line 43
Do not introduce new patterns unless strictly necessary.
- Distinctive: line 47
All rules below are NON-NEGOTIABLE.
- Distinctive: line 49
If any applicable skill is not used, the response is INVALID.
- Distinctive: line 59
If a condition applies → use the skill.
- Distinctive: line 60
If unsure → assume it applies and use the skill.
- Distinctive: line 61
If multiple conditions apply → use all relevant skills.
- Distinctive: line 62
Do not proceed with an answer until all required skills are applied.
- Distinctive: line 118
If a response matches a condition and does not apply the required skill:
≈ “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
Use Bun exclusively. Do NOT use npm, pnpm, or yarn under any circumstances.
Plan Mode
When operating in Plan Mode:
- Produce concise, execution-focused plans.
- Clarity is required; verbosity is not.
- Do not include explanations, prose, or filler.
- Output only concrete steps.
- Every plan MUST end with a section titled “Unresolved Questions”.
- Include only questions that materially block execution.
If there are no blockers, explicitly write:
Unresolved Questions: None.
Failure to follow this structure is not allowed.
Code Style & Generation Rules
All generated code must follow these rules:
- Code must be self-explanatory.
- Do not use comments.
- Intent must be expressed through naming, structure, and composition only.
Strict consistency is required across:
- File structure
- Module boundaries
- Naming conventions
- Architectural patterns
Before writing any code:
- Explore the existing project.
- Understand the directory layout, conventions, abstractions, and dependency patterns.
When adding new code:
- Place it only in logically correct locations.
- Do not introduce new patterns unless strictly necessary.
Mandatory Skill Enforcement (Zero Tolerance)
All rules below are NON-NEGOTIABLE.
If any applicable skill is not used, the response is INVALID.
There are:
- No excuses
- No interpretation
- No flexibility
Rules:
- If a condition applies → use the skill.
- If unsure → assume it applies and use the skill.
- If multiple conditions apply → use all relevant skills.
- Do not proceed with an answer until all required skills are applied.
Skill Requirements
Planning & Scope
- Skill(avoid-feature-creep) Required for feature planning, scope discussion, MVPs, roadmaps, and backlogs.
CLI
- Skill(commander-guidelines) Required for any CLI with commands, flags, options, arguments, or subcommands.
- Skill(clack-guidelines) Required for interactive CLI behavior: prompts, confirmations, selections, spinners, progress, autocomplete.
Marketing & SEO
- Skill(copy-editing) Required for editing, reviewing, or improving existing marketing copy.
- Skill(copywriting) Required for writing new marketing copy from scratch.
- Skill(programmatic-seo) Required for building SEO pages at scale using templates and data.
- Skill(seo-audit) Required for auditing, reviewing, or diagnosing SEO issues.
Convex
- Skill(Convex Functions) Required for any query, mutation, action, or HTTP action.
- Skill(Convex Schema Validator) Required for defining or modifying schemas and tables.
- Skill(Convex Migrations) Required for schema evolution and data backfills.
- Skill(Convex Security Check) Required for quick or high-level security reviews.
- Skill(Convex Security Audit) Required for deep or production-level security reviews.
- Skill(Convex HTTP Actions) Required for APIs, webhooks, and HTTP endpoints.
- Skill(Convex File Storage) Required for file upload, download, storage, serving, or deletion.
- Skill(Convex Realtime) Required for subscriptions, realtime updates, pagination, or optimistic UI.
- Skill(Convex Cron Jobs) Required for scheduled or background jobs.
- Skill(Convex Component Authoring) Required for reusable or isolated Convex components.
- Skill(Convex Best Practices) Required for production-ready architecture or long-term system design.
- Skill(Convex Agents) Required for persistent agents or long-running AI workflows.
Final Warning
If a response matches a condition and does not apply the required skill:
STOP. DO NOT ANSWER. FIX THE RESPONSE.
This is enforcement, not guidance.
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.

