Open-source, secure environment with real-world tools for enterprise-grade agents.
Python agentaiai-agentai-agentscode-interpretercopilotdevelopmentdevtoolsgptgpt-4javascriptllmnextjsopenaipythonreactsoftwaretypescript
Cleanup
Tomas Valenta511f0b0ded31
diff --git a/.vercelignore b/.vercelignoreindex 9b0f4ff50..956b47d1f 100644--- a/.vercelignore+++ b/.vercelignore@@ -1 +1,15 @@-api-service\ No newline at end of file+api-service+supabase+.github+.devcontainer+README.md+CONTRIBUTING.md+DEVELOPMENT_SETUP.md+LICENSE+Dockerfile+docker-compose.yml+.env.example+.vscode+docs-assets+.python-version+.dockerignorediff --git a/SMOL_DEPLOYMENT.md b/SMOL_DEPLOYMENT.mddeleted file mode 100644index 1c5049f4c..000000000--- a/SMOL_DEPLOYMENT.md+++ /dev/null@@ -1,20 +0,0 @@-# How the smol agent specific deployment works-- Deploy button in repo opens page `/agent/smol-developer`-- On this page user signs in with GitHub- - We need to fix GH Apps vs OAuth Apps for the sign in - we need GH App access token to retrieve repos that were given permissions by the user-- User selects repo to deploy- - They can also modify permissions for our GH App if we don't have access to repo they want to use-- User adds initial prompt that specifies what should smol dev do in the first iteration-- User specifies their OpenAI key and clicks on "Deploy"-- Then we create a PR throught the GH API (using the GH App token)- - We need to create branch and (empty) commit first to be able to create PR- - (For now we can probably create the branch directly in the repo without forking it - we need repo write access for that)-- We then create agent deployment for the specific repo- - The necessary GH data are in the `auth` field in the deployments table- - The OpenAI key is included in the `config` field alongside the model configuration- - We also add informations about the PR that was created so we can update the branch connected to the deployment via git pushes in the agent dev environemnt-- During each agent run we authenticate with GH App token, pull the repo, make the changes necessary and then push the changes to the branch connected to the deployment- - To pull and push we use the `git` command line tool but we need to authenticate with GH App token- - We can use the `git credential` helper to do that - ideally using the `gh` CLI tool-- On each comment to the PR we trigger a webhook that compiles the prompt from the PR description and comments and then triggers a new run of the smol agent- - If the smol dev agent is already processing previous PR change we cancel that run and then start a new one with the latest changesdiff --git a/package.json b/package.jsonindex bfcb101ac..9ea9558bf 100644--- a/package.json+++ b/package.json@@ -1,11 +1,11 @@{"name": "e2b",- "version": "0.1.0",- "description": "Your own virtual developer. e2b lets you build & deploy specialized AI agents that build software for you based on your instructions",+ "version": "0.2.0",+ "description": "Developer-first AgentOps platform - deploy, test, and monitor AI agents","homepage": "https://e2b.dev","author": {"name": "FoundryLabs, Inc.",- "email": "[email protected]",+ "email": "[email protected]","url": "https://e2b.dev"},"bugs": "https://github.com/e2b-dev/e2b/issues",@@ -125,4 +125,4 @@"react-dom": "$react-dom"}}-}+}\ No newline at end of file