agents-md-generator
Skill0 starsChanged 3 months agoMITView on GitHub
Generate or update an AGENTS.md / CLAUDE.md / context file for a repo. Use when the user says "set up CLAUDE.md", "generate AGENTS.md", "write a context file for this repo", "document this codebase for AI agents", or when a fresh repo has none. Also use when an existing context file has gone stale and needs a rewrite from current source. Do NOT use for editing project-specific README or contributor docs.
- Reads credentials
7 other projects use a version of this.
Adopted by
Most recently taken up by
- n-n-code/n-n-code-skillstheir own version, first seen Nov 1, 2026
- zhengwenxuan1991/skillstheir own version, first seen Sep 19, 2026
- Eriemon/agents-md-generatortheir own version, first seen Aug 26, 2026
- thienanblog/awesome-ai-agent-skillstheir own version, first seen May 3, 2026
- MADTeacher/mad-agents-skillstheir own version, first seen May 1, 2026
- jleivo/skillstheir own version, first seen Mar 9, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 7
- MADTeacher/mad-agents-skills★ 106same name
- thienanblog/awesome-ai-agent-skills★ 68same name
- Eriemon/agents-md-generator★ 14same name
- n-n-code/n-n-code-skills★ 4same name
- jleivo/skills★ 1same name
- zhengwenxuan1991/skills★ 1same name
- uts58/my-random-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository11 other files it gives its agents.
- adversarial-diff-verifySkill
- agent-harnessSkill
- plugin-manifest-referenceSkill
- publish-postSkill
- research-sweepSkill
- workflow-designSkill
- workflow-feature-specSkill
- workflow-monthly-skill-auditSkill
- workflow-stage-draftSkill
- workflow-vault-weeklySkill
- work-trackingSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
6
Distinctive
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 16
Read each language's README if present
- Distinctive: line 17
Read the root README
- Distinctive: line 52
Default behavior when config is absent
- Distinctive: line 56
Test card numbers, API keys, or credentials needed for sandbox testing
- Distinctive: line 61
If implementations make direct HTTP calls to an external API (rather than using an SDK), capture the request shape:
- Distinctive: line 80
Use this structure, targeting 80–120 lines:
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill: agents-md-generator
Produces a concise, integrator-focused AGENTS.md for repos that demonstrate one concept across multiple language implementations.
Steps
1. Explore the repo structure
- List root-level files and directories
- Identify the language implementation folders (e.g.
php/,java/,dotnet/) - Read each language's README if present
- Read the root README
2. Extract the core concept
Answer: What is the single thing this repo teaches? Distill to one sentence: "[verb] [what] using [how]".
3. Find the critical patterns
Look for:
- Non-obvious SDK/API choices that have consequences if wrong (e.g. using
VerifyvsCharge) - Flags or parameters that are required but easy to miss
- Behaviors that activate automatically (e.g. mock mode)
- Files that are canonical reference implementations
For each pattern, capture the why — especially if getting it wrong has a real cost (penalty, security issue, data loss). This is the highest-value content in the file.
4. Map the files
For each language, identify:
- The core utility/helper file (canonical implementation)
- The endpoint/controller files
- The storage layer
- The mock/test helpers
- The frontend entry point (if present)
Note specific line ranges for the most important methods (SDK config, core pattern implementation).
5. Extract the API surface
List all REST endpoints with method, path, and one-line purpose. All languages should expose identical endpoints — confirm this or note divergence.
6. Capture environment and config
- Required environment variables and their purpose
- Config file names (
.env.sample,appsettings.json, etc.) - Default behavior when config is absent
7. Find test credentials
- Test card numbers, API keys, or credentials needed for sandbox testing
- Where to get real credentials (developer portal link)
8. Check for direct API calls
If implementations make direct HTTP calls to an external API (rather than using an SDK), capture the request shape:
- Endpoint URL
- Required headers (especially auth headers — format, prefix, casing)
- Key body fields, particularly any with non-obvious values (e.g. numeric ISO codes instead of alpha codes)
Include this as a "## API Request Shape" section only if the request structure is non-obvious and getting it wrong would cause silent failures or hard-to-debug errors. Skip it for SDK-based repos where the library handles the wire format.
9. Note the architecture flow
Two or three bullet flows max:
- How data flows in (e.g. frontend → token → backend → SDK → stored)
- How payments/actions flow out
9. Identify security/production warnings
For demo repos: note what is NOT production-ready and why (no auth, no encryption, simplified storage, etc.).
10. Write the AGENTS.md
Use this structure, targeting 80–120 lines:
# [Project Name]
> [One-sentence summary: verb + what + how + languages]
## Critical Patterns
[2–4 numbered items. Each: bold rule, then plain-language explanation of the WHY.
Tone: informational, written for someone integrating this — not a maintainer warning.]
## Repository Structure
### [Language] ([framework/runtime])
- [`path/to/File.ext`](path/to/File.ext) — [role, line refs for key methods if useful]
...
[Repeat per language]
### Shared
- [config files, docker-compose, etc.]
## API Surface
| Method | Path | Purpose |
|--------|------|---------|
...
## Environment Variables
```bash
VAR_NAME=value # explanation
```
## [Test Cards / Sandbox Credentials] (if applicable)
| [Column] | [Column] | [Column] |
...
## API Request Shape (if applicable — direct API calls only)
- `field.name` — [what it is, why the value is non-obvious]
...
## Architecture Summary
**[Flow name]:** [step → step → step]
**[Flow name]:** [step → step → step]
## Security Notes
[1–3 sentences on what demo code omits for production.]
## SDK Versions
- [Language]: `package-name` vX.Y+
...Output Rules
- Tone: Written for a developer integrating this, not a maintainer guarding it. "The X file is the reference implementation — start there if adapting" not "CANONICAL — modify with caution."
- No code blocks: Line-number refs to actual source files instead of reproducing code.
- No redundant content: Anything derivable by reading the repo files (imports, full dependency lists) belongs in the repo, not here.
- Front-load the non-obvious: The Critical Patterns section is the reason this file exists. If a developer could figure it out from reading the README, it doesn't belong in Critical Patterns.
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.

