Skip to content
Back to search

GenerateAgents.md

Automated generation of comprehensive Agents.md for LLMs, driven by the DSPy Recursive language model implementation.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/originalankur--generateagents-md/badge.svg)](/listings/originalankur--generateagents-md)

HTML

<a href="/listings/originalankur--generateagents-md"><img src="/api/listings/originalankur--generateagents-md/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

These steps run a script from the network that findsafeskills does not scan.

1. Clone & Install

git clone https://github.com/originalankur/GenerateAgents.md
cd GenerateAgents.md
uv sync --extra dev     # installs all deps + dev tools in one step

> πŸ’‘ Don't have `uv`? Install it with `curl -LsSf https://astral.sh/uv/install.sh | sh` or see [uv docs](https://docs.astral.sh/uv/).

2. Set Your API Key

Copy the sample env file and fill in the key for your chosen provider:

cp .env.sample .env

*(Make sure the `.env` file sits directly in the root directory of the project, i.e., `GenerateAgents.md/.env`)*

You only need **one** provider key β€” whichever model you select:

| Provider | Env Variable | Get a key | |---|---|---| | Gemini | `GEMINI_API_KEY` | [Google AI Studio](https://aistudio.google.com/apikey) | | Anthropic | `ANTHROPIC_API_KEY` | [Anthropic Console](https://console.anthropic.com/) | | OpenAI | `OPENAI_API_KEY` | [OpenAI Platform](https://platform.openai.com/api-keys) |

*Note: You can use **any** of the 100+ providers supported by [LiteLLM](https://docs.litellm.ai/docs/providers) (e.g., `OLLAMA_API_BASE`, `OPENROUTER_API_KEY`) simply by exposing the environment variables LiteLLM expects.*

3. Run

# Default β€” generates AGENTS.md for a local repository (Gemini 2.5 Pro)
uv run autogenerateagentsmd /path/to/local/repo

# Analyze a public github repository using the flag
uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask

Using Specific Models

# Choose a specific model (supports ANY model natively supported by LiteLLM)
uv run autogenerateagentsmd /path/to/local/repo --model anthropic/claude-sonnet-4.6
uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask --model openai/gpt-5.2
uv run autogenerateagentsmd /path/to/local/repo --model ollama/llama3

# Connect to a local provider (like Ollama or vLLM) using custom endpoints: `ollama run llama3.2:1b`
uv run autogenerateagentsmd /path/to/local/repo --model ollama_chat/llama3.2 --api-base "http://localhost:11434" --api-key "optional-key"

# Pass just a default catalog provider name (gemini, anthropic, openai)
uv run autogenerateagentsmd /path/to/local/repo --model anthropic

# List all supported catalog models
uv run autogenerateagentsmd --list-models

Advanced Analysis

**1. Strict Style** Research suggests that broad, descriptive codebase summaries can sometimes distract LLMs and drive up token costs. The strict style combats this by giving the agent *only* what it can't easily `grep` for itself: strict constraints, undocumented quirks, and things it must *never* do.

uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask --style strict

**2. Analyze Git History** Automatically deduce anti-patterns from recently reverted commits. This powerful feature uses `git log --grep=revert` to inspect reverted commits in the repository (by default, looking back through the last 500 commits). It presents an **interactive prompt** where you can select exactly which reverting commits you want to analyze (e.g. `1,2,3` or `1-5`). It then feeds the diff patches of those selected commits into a dedicated DSPy module to extract explicit "Lessons Learned" and "Anti-Patterns", ensuring your AI agent avoids making the exact same mistakes previous human developers made.

# Works on both local and cloned GitHub repositories (defaults to scanning 500 commits)
uv run autogenerateagentsmd /path/to/local/repo --analyze-git-history

# Specify a custom limit for how far back to look for reverted commits (e.g., 1000)
uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask --style strict --analyze-git-history 1000

4. Find Your Output

The generated file will be saved under the `projects/` directory using the repository name.

| Output | Location | |---|---| | `AGENTS.md` | `./projects/<repo-name>/AGENTS.md` |

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md Β· README Β· 296 lines Β· 1 finding marked

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # πŸ€– GenerateAgents.md
  2. **Automatically generate Agents.md for any GitHub / Local Repository. Long context enabled using dspy.RLM aka Recursive Language Models.**
  3. GenerateAgents.md analyzes local or GitHub repositories using Recursive Language Models (dspy.RLM) to produce optimized AGENTS.md files. It features deep codebase exploration, Git history-based anti-pattern deduction, and multiple output styles (Strict vs. Comprehensive). Powered by LiteLLM, it supports **over 100+ LLM providers** (Gemini, Anthropic, OpenAI, Ollama, OpenRouter, etc.) out of the box.
  4. ---
  5. ## πŸš€ Quick Start
  6. ### 1. Clone & Install
  7. ```bash
  8. git clone https://github.com/originalankur/GenerateAgents.md
  9. cd GenerateAgents.md
  10. uv sync --extra dev # installs all deps + dev tools in one step
  11. ```
  12. > πŸ’‘ Don't have `uv`? Install it with `curl -LsSf https://astral.sh/uv/install.sh | sh` or see [uv docs](https://docs.astral.sh/uv/).
    • Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
    • Line 19 on GitHub
  13. ### 2. Set Your API Key
  14. Copy the sample env file and fill in the key for your chosen provider:
  15. ```bash
  16. cp .env.sample .env
  17. ```
  18. *(Make sure the `.env` file sits directly in the root directory of the project, i.e., `GenerateAgents.md/.env`)*
  19. You only need **one** provider key β€” whichever model you select:
  20. | Provider | Env Variable | Get a key |
  21. |---|---|---|
  22. | Gemini | `GEMINI_API_KEY` | [Google AI Studio](https://aistudio.google.com/apikey) |
  23. | Anthropic | `ANTHROPIC_API_KEY` | [Anthropic Console](https://console.anthropic.com/) |
  24. | OpenAI | `OPENAI_API_KEY` | [OpenAI Platform](https://platform.openai.com/api-keys) |
  25. *Note: You can use **any** of the 100+ providers supported by [LiteLLM](https://docs.litellm.ai/docs/providers) (e.g., `OLLAMA_API_BASE`, `OPENROUTER_API_KEY`) simply by exposing the environment variables LiteLLM expects.*
  26. ### 3. Run
  27. ```bash
  28. # Default β€” generates AGENTS.md for a local repository (Gemini 2.5 Pro)
  29. uv run autogenerateagentsmd /path/to/local/repo
  30. # Analyze a public github repository using the flag
  31. uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask
  32. ```
  33. #### Using Specific Models
  34. ```bash
  35. # Choose a specific model (supports ANY model natively supported by LiteLLM)
  36. uv run autogenerateagentsmd /path/to/local/repo --model anthropic/claude-sonnet-4.6
  37. uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask --model openai/gpt-5.2
  38. uv run autogenerateagentsmd /path/to/local/repo --model ollama/llama3
  39. # Connect to a local provider (like Ollama or vLLM) using custom endpoints: `ollama run llama3.2:1b`
  40. uv run autogenerateagentsmd /path/to/local/repo --model ollama_chat/llama3.2 --api-base "http://localhost:11434" --api-key "optional-key"
  41. # Pass just a default catalog provider name (gemini, anthropic, openai)
  42. uv run autogenerateagentsmd /path/to/local/repo --model anthropic
  43. # List all supported catalog models
  44. uv run autogenerateagentsmd --list-models
  45. ```
  46. #### Advanced Analysis
  47. **1. Strict Style**
  48. Research suggests that broad, descriptive codebase summaries can sometimes distract LLMs and drive up token costs. The strict style combats this by giving the agent *only* what it can't easily `grep` for itself: strict constraints, undocumented quirks, and things it must *never* do.
  49. ```bash
  50. uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask --style strict
  51. ```
  52. **2. Analyze Git History**
  53. Automatically deduce anti-patterns from recently reverted commits. This powerful feature uses `git log --grep=revert` to inspect reverted commits in the repository (by default, looking back through the last 500 commits). It presents an **interactive prompt** where you can select exactly which reverting commits you want to analyze (e.g. `1,2,3` or `1-5`). It then feeds the diff patches of those selected commits into a dedicated DSPy module to extract explicit "Lessons Learned" and "Anti-Patterns", ensuring your AI agent avoids making the exact same mistakes previous human developers made.
  54. ```bash
  55. # Works on both local and cloned GitHub repositories (defaults to scanning 500 commits)
  56. uv run autogenerateagentsmd /path/to/local/repo --analyze-git-history
  57. # Specify a custom limit for how far back to look for reverted commits (e.g., 1000)
  58. uv run autogenerateagentsmd --github-repository https://github.com/pallets/flask --style strict --analyze-git-history 1000
  59. ```
  60. ### 4. Find Your Output
  61. The generated file will be saved under the `projects/` directory using the repository name.
  62. | Output | Location |
  63. |---|---|
  64. | `AGENTS.md` | `./projects/<repo-name>/AGENTS.md` |
  65. ---
  66. ## Output Styles
  67. GenerateAgents supports two distinct styles for `AGENTS.md`, each tailored to different AI agent setups. You can toggle between them using the `--style` flag.
  68. Here are two examples generated for the `flask` repository:
  69. - **[Strict Style Example](projects/flask/AGENTS_strict.md)** (`--style strict`) - Focuses purely on coding constraints, anti-patterns, and repository quirks.
  70. - **[Comprehensive Style Example](projects/flask/AGENTS_comprehensive.md)** (`--style comprehensive`) - Includes high-level architectural overviews and explanations alongside constraints.
  71. ##### 1. Comprehensive Style (Default)
  72. This builds a detailed, expansive guide. It extracts high-level abstractions like project architecture, directory mappings, data flow principles, and agent personas. Great for giving a brand-new AI agent a complete tour of the repository.
  73. **Output Format:**
  74. ```markdown
  75. # AGENTS.md β€” <repo-name>
  76. ## Project Overview
  77. ## Agent Persona
  78. ## Tech Stack
  79. ## Architecture
  80. ## Code Style
  81. ## Anti-Patterns & Restrictions
  82. ## Database & State Management
  83. ## Error Handling & Logging
  84. ## Testing Commands
  85. ## Testing Guidelines
  86. ## Security & Compliance
  87. ## Dependencies & Environment
  88. ## PR & Git Rules
  89. ## Documentation Standards
  90. ## Common Patterns
  91. ## Agent Workflow / SOP
  92. ## Few-Shot Examples
  93. ```
  94. ##### 2. Strict Style
  95. Research suggests that broad, descriptive codebase summaries can sometimes distract LLMs and drive up token costs. The strict style combats this by giving the agent *only* what it can't easily `grep` for itself: strict constraints, undocumented quirks, and things it must *never* do.
  96. **Output Format:**
  97. ```markdown
  98. # AGENTS.md β€” <repo-name>
  99. ## Code Style & Strict Rules
  100. ## Anti-Patterns & Restrictions
  101. ## Security & Compliance
  102. ## Lessons Learned (Past Failures)
  103. ## Repository Quirks & Gotchas
  104. ## Execution Commands
  105. ```
  106. ---
  107. ## Developer Notes
  108. ### ✨ How It Works
  109. ```text
  110. β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
  111. β”‚ GenerateAgents Pipeline β”‚
  112. β”‚ β”‚
  113. β”‚ GitHub Repo URL β”‚
  114. β”‚ β”‚ β”‚
  115. β”‚ β–Ό β”‚
  116. β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚
  117. β”‚ β”‚ Clone │───▢│ Load Source Tree (nested dict) β”‚ β”‚
  118. β”‚ β”‚ (git) β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚
  119. β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β”‚
  120. β”‚ β–Ό β”‚
  121. β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚
  122. β”‚ β”‚ CodebaseConventionExtractor β”‚ β”‚
  123. β”‚ β”‚ β”‚ β”‚
  124. β”‚ β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ β”‚
  125. β”‚ β”‚ β”‚ ExtractCodebaseInfo (RLM Pass) β”‚ β”‚ β”‚
  126. β”‚ β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β”‚
  127. β”‚ β”‚ β–Ό β”‚ β”‚
  128. β”‚ β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ β”‚
  129. β”‚ β”‚ β”‚ CompileConventionsMarkdown (CoT) β”‚ β”‚ β”‚
  130. β”‚ β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β”‚
  131. β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚
  132. β”‚ β–Ό β”‚
  133. β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚
  134. β”‚ β”‚ AgentsMdCreator β”‚ β”‚
  135. β”‚ β”‚ β”‚ β”‚
  136. β”‚ β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ β”‚
  137. β”‚ β”‚ β”‚ ExtractAgentsSections (CoT) β”‚ β”‚ β”‚
  138. β”‚ β”‚ β”‚ (Extracts 17 specific sections) β”‚ β”‚ β”‚
  139. β”‚ β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β”‚
  140. β”‚ β”‚ β–Ό β”‚ β”‚
  141. β”‚ β”‚ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ β”‚
  142. β”‚ β”‚ β”‚ compile_agents_md() (Python) β”‚ β”‚ β”‚
  143. β”‚ β”‚ β”‚ (Template matching into markdown) β”‚ β”‚ β”‚
  144. β”‚ β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β”‚
  145. β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚
  146. β”‚ β–Ό β”‚
  147. β”‚ projects/<repo-name>/AGENTS.md β”‚
  148. β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
  149. ```
  150. ### πŸ“ Project Structure
  151. ```text
  152. GenerateAgents/
  153. β”œβ”€β”€ src/
  154. β”‚ └── autogenerateagentsmd/ # Core package directory
  155. β”‚ β”œβ”€β”€ cli.py # CLI entry point β€” orchestrates the analysis pipeline
  156. β”‚ β”œβ”€β”€ model_config.py # Provider registry, model catalog, and CLI argument parsing
  157. β”‚ β”œβ”€β”€ signatures.py # DSPy Signatures (LM task definitions)
  158. β”‚ β”‚ β”œβ”€β”€ ExtractCodebaseInfo # RLM: Extracts comprehensive codebase properties
  159. β”‚ β”‚ β”œβ”€β”€ CompileConventionsMarkdown # CoT: Compiles RLM output into markdown
  160. β”‚ β”‚ └── ExtractAgentsSections # CoT: Translates conventions -> 17 AGENTS.md fields
  161. β”‚ β”œβ”€β”€ modules.py # DSPy Modules (pipeline components)
  162. β”‚ β”‚ β”œβ”€β”€ CodebaseConventionExtractor # Performs RLM extraction & markdown compilation
  163. β”‚ β”‚ └── AgentsMdCreator # Splits info & formats final AGENTS.md text
  164. β”‚ └── utils.py # Utility functions
  165. β”‚ β”œβ”€β”€ clone_repo() # Shallow git clone
  166. β”‚ β”œβ”€β”€ load_source_tree() # Recursively map directories to a nested dict
  167. β”‚ β”œβ”€β”€ compile_agents_md() # Combines the 17 extracted fields into AGENTS.md
  168. β”‚ └── save_agents_to_disk() # Saves output to `projects/<repo_name>/`
  169. β”œβ”€β”€ tests/
  170. β”‚ └── ... # Pytest test suite, executing end-to-end tests
  171. β”œβ”€β”€ pyproject.toml # Project metadata, dependencies & tool config
  172. β”œβ”€β”€ uv.lock # Reproducible dependency lock file
  173. β”œβ”€β”€ .env.sample # Template for API keys
  174. └── .env # Your API keys (not committed)
  175. ```
  176. ---
  177. ### Environment Variables
  178. | Variable | Required | Description |
  179. `|---|---|---|
  180. | `AUTOSKILL_MODEL` | No | Default model string (avoids `--model` flag) |
  181. | `GITHUB_REPO_URL` | No | Target repository URL (skips prompt) |
  182. ### Supported Models
  183. GenerateAgents natively supports **ANY model supported by LiteLLM**!
  184. **Important:** When passing a model string to the `--model` flag, you *must* explicitly provide it in the format `PROVIDER/MODEL_NAME` (e.g., `ollama/llama3`, `openrouter/anthropic/claude-3-opus`, `openai/gpt-4o`). This ensures LiteLLM appropriately maps the request to the correct provider API. Make sure the corresponding required environment variables are exported.
  185. You can also pass custom endpoints and API keys directly via the CLI, which is especially useful for local providers:
  186. * `--api-base`: Allows pointing to custom local LM endpoints (e.g., `http://localhost:11434` for Ollama).
  187. * `--api-key`: Explicitly pass an API key rather than relying on environment variables.
  188. For ease of use, we maintain a "Catalog" of default fallbacks. If you just pass the provider name, these are the models we use by default:
  189. | Provider | Default Model |
  190. |---|---|
  191. | Gemini | `gemini/gemini-2.5-pro` |
  192. | Anthropic | `anthropic/claude-sonnet-4.6` |
  193. | OpenAI | `openai/gpt-5.2` |
  194. Run `uv run autogenerateagentsmd --list-models` to view our catalog and defaults.
  195. ---
  196. ### πŸ§ͺ Testing
  197. The project includes a robust test suite covering fast unit tests, git integration tests, and full end-to-end pipeline executions using real LLM API calls.
  198. #### Test Architecture Overview
  199. ```text
  200. tests/
  201. β”œβ”€β”€ conftest.py # Session-scoped fixtures (model setup, DSPy config)
  202. β”œβ”€β”€ test_cli.py # Unit tests for CLI parsing (fully mocked)
  203. β”œβ”€β”€ test_model_config.py # Unit tests for model configuration & catalog
  204. β”œβ”€β”€ test_utils.py # Unit tests for utility functions
  205. β”œβ”€β”€ test_modules.py # Unit tests for DSPy module initialization
  206. β”œβ”€β”€ test_analyze_git_history.py # Integration tests using a temporary local git repo
  207. β”œβ”€β”€ test_e2e_pipeline.py # Full end-to-end pipeline tests (real API calls)
  208. └── output/ # Persistent output from E2E test runs (ignored in git)
  209. ```
  210. #### Running Tests
  211. Tests use `pytest`. The end-to-end tests require a valid API key configured in `.env` and internet access, as they clone real GitHub repositories and make real LLM calls. The E2E tests are marked with `@pytest.mark.e2e`.
  212. ```bash
  213. # Run ALL tests (unit, integration, and E2E)
  214. uv run pytest tests/ -v -s
  215. # Run ONLY E2E tests
  216. uv run pytest tests/ -v -s -m e2e
  217. # Run only unit/integration tests (fast, no API hits required)
  218. uv run pytest tests/ -v -s -m "not e2e"
  219. # Test with a specific LLM provider for E2E
  220. AUTOSKILL_MODEL=openai/gpt-5.2 uv run pytest tests/ -v -s -m e2e
  221. # Run tests filtering by a specific keyword
  222. uv run pytest tests/ -v -s -k "test_clone"
  223. ```
  224. > ⚠️ **Note:** E2E pipeline tests make real LLM API calls and may take 1-2 minutes per test. Test generated `AGENTS.md` and markdown outputs are persisted in `tests/output/<repo_name>/` for manual inspection.
  225. ---
  226. ## πŸ“œ License
  227. [MIT](LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.