Free lite scanner + paid $7 repo safety bundle
AI coding agents can read instructions, run package scripts, touch MCP/Cursor/Claude configs, and trip over secret-adjacent files. This starter pack gives you a cheap first pass before Claude Code, Cursor, Codex-style agents, or local automations start working.
Using microsoft/playwright-mcp to let Claude Code, Cursor, Codex, or another agent drive a browser session? This buyer-specific receipt shows what to check before logged-in apps, localhost services, form submissions, package scripts, or secret-adjacent config enter scope. PLAYWRIGHT_MCP_BROWSER_AGENT_PROOF
Open the Playwright MCP browser-agent proof Get the reusable $7 workflow
Using a2aproject/A2A or a similar coding-agent workflow? This buyer-specific receipt shows what to check before agent-to-agent connectors, package scripts, API tools, workflow files, and secret-adjacent config enter scope. A2A_AGENT_PROTOCOL_WORKFLOW_PROOF
Open the Agent2Agent protocol proof Get the reusable $7 workflow
Run the free scan first. The paid pack is for the moment the scan creates real handoff work — not for fear-buying a blank repo.
Two or more risk buckets, package scripts, MCP/Cursor/Claude config, secret-adjacent files, or shell commands the agent might run. The bundle gives you the reusable report template, destructive-command hook, demo, and verifier instead of rebuilding them during the task.
Zero or one low-risk signal, no agent config, no risky install/deploy scripts, and no secret-adjacent files. Use the free lite scanner and normal code review discipline.
Using JetBrains/koog-style Koog JVM/Kotlin agents? This buyer-specific receipt shows what to check before Gradle/Maven scripts, workflow permissions, API tools, deployment config, generated edits, or secret-adjacent env values enter scope. JETBRAINS_KOOG_JVM_AGENT_PROOF
Using mistralai/mistral-vibe or a Vibe-style terminal coding assistant? This buyer-specific receipt shows what to check before repo context, shell/package commands, workflow files, MCP/tool connectors, or secret-adjacent config enter scope.
Open the Mistral Vibe coding assistant proof Get the reusable $7 workflow
Using camel-ai/camel or a similar agent framework? This buyer-specific receipt shows what to review before multi-agent role/task prompts, tool/API connectors, package scripts, generated repo edits, shell commands, and .env-adjacent values enter an agent run. CAMEL_AI_MULTI_AGENT_PROOF
Open the CAMEL-AI preflight proof Get the reusable $7 workflow
Using google/adk-python or a Google ADK-style Python agent workflow? This buyer-specific receipt shows what to review before repo files, package scripts, tool/API connectors, deployment config, or secret-adjacent environment values enter the agent run. GOOGLE_ADK_PYTHON_AGENT_PROOF
Using PrefectHQ/fastmcp or a FastMCP-style Python MCP server? This buyer-specific receipt shows what to review before MCP tools, package commands, API calls, env values, or shell access enter a Claude/Cursor/Codex agent session. FASTMCP_TOOL_SERVER_PROOF
Open the FastMCP preflight proof Get the reusable $7 workflow
Using aws/amazon-q-developer-cli or an Amazon Q-style terminal AI coding assistant? This buyer-specific receipt shows what to review before repo edits, shell commands, package/deploy scripts, or AWS credential-adjacent files enter scope. AMAZON_Q_DEVELOPER_CLI_PROOF
Open the Amazon Q preflight proof Get the reusable $7 workflow
Using charmbracelet/crush or a Crush-style terminal coding agent? This buyer-specific receipt shows what to review before repo edits, package scripts, workflow files, shell commands, or secret-adjacent paths enter the task. CRUSH_TERMINAL_AGENT_PROOF
Using Cline in VS Code with workspace files, terminal commands, MCP/browser tools, package scripts, or secret-adjacent files? This buyer-specific receipt shows what to freeze before the agent run and when a Yellow/Red scan should trigger the reusable $7 workflow.
Open the Cline preflight proof Get the reusable $7 workflow
git clone https://github.com/el-zachariah/ai-agent-safety-starter-pack.git
cd ai-agent-safety-starter-pack
python3 agent_preflight_lite.py /path/to/repoThe lite scanner flags agent instruction files, MCP / Claude / Cursor / Cline / GitHub Copilot instruction config, workflow files, risky shell patterns, package scripts, and secret-adjacent filenames worth reviewing before a run.
Using anthropics/claude-code-action in GitHub Actions? This buyer-specific receipt shows what to review before the workflow comments on PRs, edits files, runs package scripts, or receives secret-adjacent CI context. CLAUDE_CODE_ACTION_GITHUB_WORKFLOW_PROOF
Open the Claude Code Action preflight proof Get the reusable $7 workflow
Using CodebuffAI/codebuff or a Codebuff-style terminal coding agent? This buyer-specific receipt shows what to review before repo edits, package scripts, workflow files, shell commands, or secret-adjacent paths enter the task. CODEBUFF_TERMINAL_AGENT_PROOF
Open the Codebuff preflight proof Get the reusable $7 workflow
Using agno-agi/agno or an Agno-style agent framework? This buyer-specific receipt shows what to review before repo files, package scripts, MCP/API connectors, vector/database config, or secret-adjacent environment values enter the agent run. AGNO_AGENT_FRAMEWORK_PROOF
Using Kilo-Org/kilocode or a similar VS Code agentic-engineering extension? This buyer-specific receipt shows what to review before workspace context, terminal commands, MCP/tool servers, package scripts, or secret-adjacent files enter scope. KILO_CODE_AGENTIC_ENGINEERING_PROOF
Open the Kilo Code preflight proof Get the reusable $7 workflow
Using cloudflare/agents, Workers AI, Durable Objects, or wrangler deploy flows? This buyer-specific receipt shows what to review before package scripts, .dev.vars, API-backed tools, shell commands, or deploy config enter an agent run.
Open the Cloudflare Agent preflight proof Get the reusable $7 workflow
Using zed-industries/zed, Zed Agent Panel, or editor-native AI coding workflows? This buyer-specific receipt shows what to check before repo edits, package scripts, MCP connectors, shell commands, or secret-adjacent files enter scope.
Using bmad-code-org/BMAD-METHOD or BMAD-style analyst/PM/architect/dev/QA agents? This buyer-specific receipt shows what to check before generated stories, package scripts, repo edits, or automation credentials enter an agent run.
Using stackblitz-labs/bolt.diy, Bolt.new-style builders, or hosted browser AI app builders? This buyer-specific receipt shows what to review before generated package scripts, .env files, preview servers, or deploy config enter the agent run.
Open the Bolt-style preflight proof Get the reusable $7 workflow
python3 agent_preflight_lite.py examples/sample-risky-repoYou should see five demo findings: an agent instruction file, a secret-adjacent file, package scripts, a curl-piped shell install, and a destructive delete command. That gives you a quick proof of what the scanner will surface in your own repo.
The scorecard converts the free scan into a fast Green / Yellow / Red decision: run normally, add a handoff note, or stop and set guardrails before the agent gets shell access.
Buyer trigger: if the scan lands Yellow or Red and the agent will run commands in a real repo, the $7 bundle saves the setup work of building the report template, hook, and verifier yourself.
Open the one-minute scorecard Buy the $7 starter pack
Not sure what to do after the scan flags `AGENTS.md`, MCP config, package scripts, secret-adjacent files, or risky shell patterns? Use the free matrix to turn findings into run / ask / avoid decisions before the agent gets shell access.
Open the red-flag matrix Buy the $7 starter pack
Copy the free /agent-preflight command into a repo before a tool-enabled Claude Code session. It tells the agent to run the lite scanner, produce a Green / Yellow / Red decision, and write the run / ask / avoid handoff before shell commands begin.
Install the free slash command Buy the $7 starter pack
Running a CrewAI-style crew with local tools, package scripts, .env values, or write-capable API calls? This buyer-specific receipt shows what to review before multiple agents inherit repo and token scope.
Open the CrewAI preflight proof Get the reusable $7 workflow
Running AutoGen-style planner/executor agents with function calls, code execution, MCP/browser tools, package scripts, or API-backed actions? This buyer-specific receipt shows what to freeze before multiple agents inherit repo and token scope.
Open the AutoGen preflight proof Get the reusable $7 workflow
Running OpenAI Agents SDK workflows with handoffs, function tools, MCP/API connectors, package scripts, or environment-backed execution? This buyer-specific receipt shows what to review before agents inherit repo and credential scope.
Open the OpenAI Agents SDK preflight proof Get the reusable $7 workflow
Using GitHub Copilot coding agent with repository instructions, workflow secrets, issue/PR automation, or package scripts? This buyer-specific receipt shows what to freeze before Copilot gets a task and when the Yellow/Red scan should trigger the reusable $7 workflow.
Open the Copilot coding agent proof Get the reusable $7 workflow
Using PR-Agent or another PR review bot with issue comments, workflow context, package scripts, or repository permissions? This buyer-specific receipt shows what to freeze before the bot gets trusted automation scope.
Open the PR-Agent review bot proof Get the reusable $7 workflow
Running Hugging Face smolagents, a CodeAgent, or a Python tool-calling workflow with repo files, package scripts, browser/API tools, or local environment scope? This buyer-specific receipt shows what to freeze before the agent gets tool access.
Open the smolagents preflight proof Get the reusable $7 workflow
Leaving a persistent Claude Code, MCP, or hook loop attached to a repo? This customer-specific example shows the 5-minute receipt to write before long-running state, shell commands, and tool access begin.
Open the Continuous Claude-style preflight Get the reusable $7 workflow
Trying agents or plugins from a Claude Code, Codex CLI, Cursor, OpenCode, Copilot, or Gemini marketplace? This buyer-specific receipt shows what to check before a package inherits repo instructions, MCP/tool configs, and shell access.
Open the marketplace-agent preflight Get the reusable $7 workflow
Reviewing a third-party agent/plugin/skill marketplace PR before listing it? This proof gives maintainers a concrete 5-minute receipt for commands, package scripts, MCP/tool config, secret-adjacent files, and shell scope before downstream users install the submission.
Open the marketplace PR reviewer proof Get the reusable $7 workflow
Evaluating a runtime safety wrapper or hook layer for Claude Code/Codex-style agents? Run this free preflight first so the safety layer inherits reviewed repo instructions, MCP/tool configs, package scripts, secret-adjacent files, and shell rules.
Open the runtime safety hook preflight Get the reusable $7 workflow
Running a persistent Claude Code, MCP, or agent harness across sessions? Use the free 7-minute receipt example to decide what the agent may run, what it must ask about, and when the $7 workflow bundle saves setup time before long-running state inherits repo risk.
Open the continuous-agent example Buy the $7 starter pack
Using Claude Code hooks, TDD guardrails, MCP routing, or a continuous agent loop? The free proof page shows a 6-minute preflight receipt for exactly what the agent may run, what it must ask about, and when the paid bundle saves repeated setup.
Open the hook preflight proof Buy the $7 starter pack
Adding a filesystem, GitHub, database, or internal-tools MCP server to Claude Code or Cursor? This customer-specific proof shows the Yellow handoff to write before the agent can read tokens, call tools, or run shell commands.
Open the MCP config receipt example Get the reusable $7 workflow
Need a visible receipt before an AI agent works on a pull request or issue? Paste the free review comment template with the scan decision, risk buckets, allowed commands, must-ask actions, and the exact Yellow/Red upgrade trigger.
Open the review comment template Buy the $7 starter pack
For GitHub maintainers, plugin directory curators, and small teams reviewing AI-agent PRs: this proof page shows the exact receipt to ask for before Claude Code, Codex, Cursor, or an MCP-backed agent gets shell access.
Open the maintainer receipt example Get the reusable $7 workflow
Turn the scan into a short operating note before Claude Code, Codex, Cursor, or an MCP-backed agent gets shell access. The playbook gives you a 10-minute scan → triage → run/avoid-list flow, plus copy-paste guardrail text for the agent task.
Read the free handoff playbook
Need a faster handoff? Use the free prompt block to tell Claude Code, Codex, Cursor, or another coding agent what it may run, what it must ask about first, and when the $7 bundle saves setup time.
Open the copy-paste task prompt Get the full $7 workflow
If the free scan finds agent configs, package hooks, risky shell patterns, or secret-adjacent files, use this checklist to decide whether the $7 bundle saves enough setup time to justify buying now.
Open the upgrade decision checklist Buy the $7 starter pack
Using Aider or a similar terminal AI pair-programmer in a real checkout? This customer-specific proof shows the receipt to write before the assistant edits files, runs tests, or touches install/build scripts.
Open the Aider preflight receipt Get the reusable $7 workflow
Using SWE-agent or a similar issue-to-patch autonomous coding agent in a real checkout? This customer-specific proof shows the receipt to write before the agent edits files, runs tests, or touches package/workflow scripts.
Open the SWE-agent preflight receipt Get the reusable $7 workflow
Using Continue.dev or a similar IDE coding agent with workspace context and terminal access? This customer-specific proof shows the 5-minute receipt before .continue/ configs, MCP/context providers, package scripts, and token boundaries are exposed.
Open the Continue.dev IDE agent preflight Get the reusable $7 workflow
Running Ruflo or a Claude Flow-style multi-agent swarm with Claude Code, Codex, Hermes, MCP tools, shell commands, memory, or package scripts? This buyer-specific receipt shows what to freeze before the swarm fans out work.
Open the Ruflo swarm preflight receipt Get the reusable $7 workflow
Using Composio-style toolkits, MCP connectors, auth-backed actions, or sandboxed workbenches with an AI agent? This customer-specific receipt shows what to freeze before repo context, API tools, package scripts, or credential-adjacent files enter the run.
Open the Composio tool-agent preflight receipt Get the reusable $7 workflow
Use the public scanner to surface instruction files, agent configs, secret-adjacent filenames, package scripts, workflows, and risky shell patterns before a tool-enabled agent run.
The paid starter pack gives you the reusable preflight mini, destructive-command hook, report templates, buyer quickstart, and local verifier so you do not rebuild the safety workflow from scratch.
Drop the report into your handoff notes before assigning Claude Code, Codex, Cursor, or another AI coding agent to install, test, refactor, or deploy.
Developers and small teams letting AI agents work inside real repositories, especially unfamiliar repos with install, test, deploy, or automation scripts.
Full repo preflight mini, destructive-command hook, buyer quickstart, demo risky repo/report, report templates, verification scripts, and limitations FAQ.
Not a sandbox, malware scanner, secret scanner, or full security audit. It is a fast practical checklist before an agent gets tools.
Get the full $7 bundle View a sample report
The lite scanner, docs, examples, commit history, and issue tracker are public on GitHub before you pay.
The paid ZIP is designed to run locally; you do not send your private repo, secrets, or code to Signal Loom Works.
Use GitHub Issues for pre-sale questions or install friction. Do not post private code or secrets.
Read trust + support notes Buy the $7 starter pack
Curating Claude Code plugins or AI-agent tools? Read the concise proof page showing the exact user, checked signals, sample report, and why this free repo is a low-risk pre-tool-access utility.
Open the curator proof page Buy the $7 starter pack
New buyers do not have to trust a silent checkout page. The live proof page links open directory/marketplace PRs, free repo assets, support expectations, and the Payhip bundle so Claude Code and agent-tooling teams can confirm the seller is active.
Read the live distribution proof before choosing whether to buy the low-ticket starter pack.
Run the free local preflight before enabling an enforcement hook, then keep the Green/Yellow/Red receipt with the first hook-enforced PR.
For teams using Agent Skills / n-skills-style workflows with Claude Code, Codex, Cursor, or MCP-backed agents, the free repo now ships a portable skills/agent-safety-preflight/SKILL.md package that turns repo preflight into an installable habit before command access.
Open the Agent Skills package Get the reusable $7 workflow
For teams using openai/codex-style AI coding agents, the new Codex preflight example shows how to paste a risk receipt into the task before the agent edits, tests, or opens a PR.
Read the Codex CLI preflight example or buy the $7 full pack when the scan is Yellow/Red.
Using Google Gemini CLI or a similar terminal AI coding agent in a real checkout? This customer-specific proof shows the receipt to write before the agent edits files, runs tests, or touches package/workflow scripts.
Read the Gemini CLI preflight example or buy the $7 full pack when the scan is Yellow/Red.
Using E2B-style sandboxes or code-interpreter agents against a real repository? This customer-specific proof shows what to check before the agent receives repo files, env vars, sandbox templates, package scripts, or API-backed execution scope.
Read the E2B code-interpreter preflight example or buy the $7 full pack when the scan is Yellow or Red.
Using OpenCode or a similar terminal coding agent in a real checkout? This customer-specific proof shows the receipt to write before the agent edits files, runs tests, or touches package/workflow scripts.
Read the OpenCode preflight example or buy the $7 full pack when the scan is Yellow/Red.
For teams using Cline, Roo Code, or similar editor agents, the new VS Code agent extension proof shows how to paste a risk receipt into the agent chat before the extension edits files, touches MCP/tool configs, or runs terminal commands.
Read the VS Code agent extension preflight example or buy the $7 full pack when the scan is Yellow/Red.
Using Cursor with .cursorrules, .cursor/rules/*, background agents, or repo-level rule packs? This proof shows the receipt to write before Cursor inherits project instructions, MCP/tool config, package scripts, or broad edit/test scope.
Read the Cursor rules preflight example or buy the $7 full pack when the scan is Yellow/Red.
For teams using Block Goose, MCP extensions, or shell-enabled local agents before granting repo, env, browser, or package-script scope.
Read the Goose local-agent proof or buy the $7 pack when the scan is Yellow/Red.
Running LangGraph-style stateful agents with tool calls, checkpoint stores, MCP/API connectors, package scripts, or deployment context? This buyer-specific proof shows the receipt to write before the graph agent gets repo and shell scope.
Open the LangGraph preflight proof Get the reusable $7 workflow
Using OpenHands-style autonomous software-engineering agents to open task branches, run shell commands, and change real repos? This buyer-specific proof shows the receipt to write before the agent gets repo and package-script scope.
Open the OpenHands preflight proof Get the reusable $7 workflow
Building typed Python agents with Pydantic AI, toolsets, MCP/API connectors, environment-backed models, or package scripts? This buyer-specific receipt shows what to freeze before a production agent run gets repo and credential scope.
Open the Pydantic AI preflight proof Get the reusable $7 workflow
Building Mastra-style TypeScript agents with tools, memory/storage config, package scripts, MCP/API connectors, or deployment secrets? This buyer-specific receipt shows what to freeze before the agent gets repo and shell scope.
Open the Mastra preflight proof Get the reusable $7 workflow
Running Claude Code, Codex, Cursor, Copilot coding agent, or another AI coding agent inside a Dev Container or GitHub Codespaces workspace? This buyer-specific receipt shows what to review before .devcontainer/devcontainer.json, lifecycle commands, mounts, forwarded ports, package scripts, or repo write access enter the run.
Open the Dev Container preflight proof Get the reusable $7 workflow
MCP server maintainers can verify the free preflight path first, then buy the $5 pack only when MCP/tooling changes enter a yellow or red review zone.
Building with Semantic Kernel agents, planners, plugins, memory connectors, or tool-calling workflows? This buyer-specific proof shows the receipt to write before the agent inherits repo context, package scripts, API connectors, or secret-adjacent config.
Open the Semantic Kernel preflight proof Get the reusable $7 workflow
For teams evaluating QwenLM/qwen-code / Qwen-style terminal coding agents before granting repo, shell, package-script, MCP/tool, or .env scope.
Read the Qwen Code CLI proof · Buy the $7 checklist when the scan is Yellow/Red
Buy trigger: use the paid starter pack when Cline will touch CI, deployment, package-manager, MCP, or multi-file refactor paths.
For teams using Roo Code as an in-editor agent team before granting auto-approval, terminal commands, MCP/tools, package scripts, deployment config, or credential-adjacent repo scope.
Read the Roo Code preflight proof · Buy the $7 pack if your scan is Yellow/Red
Before adding commands, hooks, skills, or MCP/tooling to a real repo, run the free preflight receipt first. If the receipt needs to become a team handoff, use the paid starter pack.
For teams using n8n, workflow credentials, webhooks, custom code nodes, or low-code AI workflows before granting an agent repo/API/package-script scope.
Buy the $7 pack when the preflight is Yellow/Red because the workflow touches credentials, webhooks, MCP/API tools, repo files, or deployment paths. Skip it when the workflow has no real credentials or repo access.
Read the n8n workflow preflight proof · Buy the $7 starter pack
For teams connecting OpenAI Agents SDK tool calls, MCP servers, package scripts, or API-backed actions before granting repo scope.
Read the OpenAI Agents SDK preflight proof · Buy the $7 remediation pack if your scan is Yellow/Red
For TypeScript/Next.js teams using Vercel AI SDK agents, tool calls, route handlers, server actions, package scripts, MCP/API connectors, or preview/deploy settings before granting repo scope.
Read the Vercel AI SDK preflight proof · Buy the $7 remediation pack if your scan is Yellow/Red
For teams letting a local interpreter read repositories, run shell/Python/package commands, or work near .env/API/browser scope before a repo preflight receipt exists.
Read the Open Interpreter preflight proof · Buy the $7 remediation pack if your scan is Yellow/Red
For high-trust Claude Code plugin and skill marketplace reviewers, the free scanner now flags .claude-plugin/ files before install and uses Yellow/Red results as the honest upgrade trigger.
Read the skills-curated marketplace proof or buy the full $7 bundle at Payhip.
For Langfuse, Phoenix, and agent eval teams: add a repo-scope preflight receipt before traced AI coding agents receive write, shell, package-script, MCP, browser, or API backed access.
Read the observability and eval proof or buy the $7 pack when the free receipt is Yellow or Red.
Using HumanLayer or another human-in-the-loop approval gate for Claude Code, Codex, OpenCode, or coding-agent actions? This proof shows the receipt to attach before approving shell commands, package scripts, MCP/API calls, or repo edits.
Open the HumanLayer approval preflight proof Get the reusable $7 workflow
Testing agentclientprotocol/agent-client-protocol or another ACP-compatible editor agent? This buyer-specific receipt shows what to freeze before the client hands repo context, package scripts, MCP/editor config, or shell scope to the coding agent.
Open the ACP agent preflight proof Get the reusable $7 workflow
For teams letting OpenHands clone, edit, run shell/package scripts, browse, or use API-backed tools inside real repos, run the free preflight first, then buy the $7 pack only when the result is Yellow/Red and a reusable approval receipt is needed.
Read the OpenHands autonomous SWE agents preflight proof · Buy the $7 pack
Aider local AI pair-programming teams now have a copy-paste preflight receipt before local AI pair-programming touches package scripts, tests, CI, or secret-adjacent files.
Read the Aider preflight proof. If the free scan is Yellow/Red, buy the $7 pack before the next run: AI Agent Safety Starter Pack.
AIDER_LOCAL_AI_PAIR_PROGRAMMING_PROOF
For teams using plandex-ai/plandex-style planning/editing agents before repo writes, shell commands, package scripts, MCP/tool config, or secret-adjacent files enter scope.
Buy trigger: Green means skip it; Yellow/Red means buy the $7 AI Agent Safety Starter Pack when you need repeatable hooks, reviewer prompts, and receipts before the agent runs.
For teams using openai/codex-style coding agents before repo writes, shell commands, package scripts, MCP/tool config, or secret-adjacent files enter scope.
Buy trigger: Green means skip it; Yellow/Red means buy the $7 AI Agent Safety Starter Pack when you need repeatable hooks, reviewer prompts, and receipts before the agent runs.
For teams using deepset-ai/haystack or Haystack-style agent/RAG workflows before repo context, tool functions, package scripts, workflow files, API connectors, or secret-adjacent paths enter scope.
Buy trigger: Green means skip; Yellow/Red means buy the $7 starter pack when the team needs repeatable hooks, reviewer prompts, and handoff receipts.
For teams using langchain-ai/langgraph or langgraph graph-agent workflows before repository context, tool calls, memory/state, generated commands, package scripts, workflow files, API connectors, or secret-adjacent paths enter scope.
Buy trigger: Green means skip; Yellow/Red means buy the $7 starter pack when the team needs repeatable hooks, reviewer prompts, and handoff receipts.
Read the LangGraph graph-agent workflow proof · Get the $7 pack
For teams using modelcontextprotocol/servers or MCP-style server/tool workflows before agent clients receive repository context, filesystem access, browser/API tools, package scripts, workflow edits, credentialed connectors, or secret-adjacent config.
Buy trigger: Green means skip; Yellow/Red means buy the $7 starter pack when the team needs repeatable hooks, reviewer prompts, and handoff receipts.
For teams using langchain-ai/langchain or LangChain-style tool/retriever/function-calling agents before repo context, generated commands, package scripts, workflow files, callbacks, MCP/API connectors, or secret-adjacent config enter scope.
Buy trigger: Green means skip; Yellow/Red means buy the $7 starter pack when the team needs repeatable hooks, reviewer prompts, and handoff receipts.
For teams about to let Microsoft AutoGen or similar agent workflows read repo context, call tools, run package scripts, or touch secret-adjacent config before a human has a receipt.
Open the Microsoft AutoGen proof · Buy the $7 checklist/templates if your scan is Yellow or Red
Named proof for OpenHands/OpenHands teams before an autonomous agent receives repo context, shell commands, workflow/package files, or secret-adjacent config.
Read the OpenHands preflight proof — Yellow/Red scans should use the $7 checklist bundle before the run.
smolagents tool-calling preflight proof shows `huggingface/smolagents` users how to run the free Green / Yellow / Red scanner before `tools.py`, `managed agents`, `API/tool credentials` enter an agent task.
Buy trigger: buy the $7 pack only when the free scan is Yellow/Red and you need the copy-paste hook, PR receipt, and handoff template today.
Running VoltAgent-style agent workflows? Use the free preflight receipt before write, shell, MCP/tool, browser, or credential-adjacent scope enters the run.
Read the VoltAgent preflight proof. Buy the $7 pack only when the free receipt is Yellow/Red and you need repeatable hooks, prompts, and handoff templates.
For teams using Significant-Gravitas/AutoGPT or AutoGPT-style autonomous workflows before repository context, tool/API connectors, package scripts, browser/API actions, workflow files, or secret-adjacent config enter scope.
Read the AutoGPT agent-platform proof · Buy the $7 starter pack when the receipt is Yellow/Red
AUTOGPT_AGENT_PLATFORM_PROOF