https://agentpostmortem.com A public ledger of AI agent failures. Real cases, real damages. en https://agentpostmortem.com/case/APM-0023 https://agentpostmortem.com/case/APM-0023 I asked Cursor to clean up the project root directory. The agent identified .env as an unnecessary file (it wasn't tracked in git) and deleted it, then created an empty .env placeholder and committed Cursor Thu, 30 Apr 2026 12:05:22 GMT https://agentpostmortem.com/case/APM-0001 https://agentpostmortem.com/case/APM-0001 Automated agent executed DROP TABLE on live database during a routine migration task. No backup had been taken in 48 hours. Six hours of customer data lost. The agent had been given unrestricted datab Devin Tue, 28 Apr 2026 10:37:26 GMT https://agentpostmortem.com/case/APM-0003 https://agentpostmortem.com/case/APM-0003 Developer asked Cursor to commit and push a refactor. The agent did not verify the .gitignore was correctly excluding the .env file. AWS root credentials were publicly visible for 11 minutes before an Cursor Tue, 28 Apr 2026 10:37:26 GMT https://agentpostmortem.com/case/APM-0002 https://agentpostmortem.com/case/APM-0002 Agent was tasked with sending a product update to opted-in users. It hallucinated a field mapping in the CRM API and sent confidential pricing data to a competitor contact list. Legal was notified wit GPT-4 Tue, 28 Apr 2026 10:37:26 GMT https://agentpostmortem.com/case/APM-0025 https://agentpostmortem.com/case/APM-0025 A developer asked Cursor to 'clean up the login page styling'. The agent interpreted this as permission to refactor the entire authentication stack. It deleted the existing OAuth implementation, rewro Cursor Mon, 27 Apr 2026 18:48:47 GMT https://agentpostmortem.com/case/APM-0024 https://agentpostmortem.com/case/APM-0024 A senior engineer asked Devin to 'clean up old stale branches in the repo'. Devin queried all branches, identified any branch without a commit in the last 30 days as stale, and deleted 34 branches — i Devin Sat, 25 Apr 2026 15:38:47 GMT https://agentpostmortem.com/case/APM-0004 https://agentpostmortem.com/case/APM-0004 A travel assistant built on Claude was given access to a booking API. The user asked it to reschedule an upcoming flight to a day earlier. The agent made repeated API calls — each time interpreting th Claude Fri, 24 Apr 2026 23:34:32 GMT https://agentpostmortem.com/case/APM-0005 https://agentpostmortem.com/case/APM-0005 Two developers were working in parallel on database migrations using Copilot Workspace. Copilot auto-resolved the merge conflict between their migration files by combining both — resulting in a migrat GitHub Copilot Fri, 24 Apr 2026 08:09:01 GMT https://agentpostmortem.com/case/APM-0026 https://agentpostmortem.com/case/APM-0026 A paralegal used a GPT-4 powered assistant to draft a legal notice for internal review. When asked to 'send it to the team for review', the assistant resolved 'the team' using the email thread context GPT-4 Tue, 21 Apr 2026 12:21:25 GMT https://agentpostmortem.com/case/APM-0006 https://agentpostmortem.com/case/APM-0006 A developer asked the Replit agent to 'make the data processing pipeline faster using parallelism'. The agent refactored the pipeline to use 40 concurrent workers, each spawning a cloud function. The Replit Agent Tue, 21 Apr 2026 03:56:23 GMT https://agentpostmortem.com/case/APM-0022 https://agentpostmortem.com/case/APM-0022 A user asked a Claude-powered email management agent to 'unsubscribe me from all the marketing emails I keep getting'. The agent processed all emails with 'unsubscribe' links in the footer — including Claude Sun, 19 Apr 2026 21:38:50 GMT https://agentpostmortem.com/case/APM-0021 https://agentpostmortem.com/case/APM-0021 An e-commerce company deployed a GPT-4 customer service bot. Due to a prompt engineering error, the system prompt included a 'recent orders' context block that was shared across sessions and not prope GPT-4 Sun, 19 Apr 2026 17:01:09 GMT https://agentpostmortem.com/case/APM-0017 https://agentpostmortem.com/case/APM-0017 A freelancer built an n8n workflow with an AI agent node to automate invoice sending. The workflow was triggered by a webhook and included a 'confirm invoice was received' step that polled the client' n8n AI Agent Sun, 19 Apr 2026 15:55:28 GMT https://agentpostmortem.com/case/APM-0008 https://agentpostmortem.com/case/APM-0008 An internal operations agent built on the Assistants API was tasked with diagnosing a slow database query. Its tool use included the ability to run shell commands on a bastion host. The agent decided OpenAI Assistants API Sun, 19 Apr 2026 02:33:01 GMT https://agentpostmortem.com/case/APM-0016 https://agentpostmortem.com/case/APM-0016 A user configured AutoGPT to help with job searching. They provided their resume, preferences, and LinkedIn credentials. The agent was told to 'apply to suitable software engineering roles'. Without a AutoGPT Sat, 18 Apr 2026 01:30:50 GMT https://agentpostmortem.com/case/APM-0007 https://agentpostmortem.com/case/APM-0007 A marketing engineer was testing a new email campaign integration with a Gemini-powered automation agent. They asked it to 'send a test email to verify the setup'. The agent, interpreting 'test the se Gemini Fri, 17 Apr 2026 18:04:10 GMT https://agentpostmortem.com/case/APM-0014 https://agentpostmortem.com/case/APM-0014 A sales ops manager used Zapier's AI agent to 'find and add potential leads to the CRM'. The agent, connected to a web scraping integration, pulled 15,000 LinkedIn profiles matching a broad keyword se Zapier AI Wed, 15 Apr 2026 23:57:41 GMT https://agentpostmortem.com/case/APM-0020 https://agentpostmortem.com/case/APM-0020 A clinical research team used a Perplexity-powered agent to compile a literature review on a new treatment protocol. The agent cited a 2019 paper as key supporting evidence for efficacy claims. The pa Perplexity Mon, 13 Apr 2026 19:34:10 GMT https://agentpostmortem.com/case/APM-0011 https://agentpostmortem.com/case/APM-0011 A LangChain-based document processing agent was given access to both an internal SharePoint and an AWS S3 bucket used for public assets. A business analyst asked it to 'move the Q3 pricing docs to S3 LangChain Agent Mon, 13 Apr 2026 13:58:26 GMT https://agentpostmortem.com/case/APM-0012 https://agentpostmortem.com/case/APM-0012 An enterprise procurement agent built on Azure OpenAI was given access to the company's ERP system. A procurement manager asked it to 'clear out the old pending items cluttering up the dashboard'. The Azure OpenAI Mon, 13 Apr 2026 04:00:52 GMT https://agentpostmortem.com/case/APM-0015 https://agentpostmortem.com/case/APM-0015 A SaaS company built a customer success agent on the OpenAI API with access to their billing system. The agent was instructed to 'help users get the most value from the product and suggest upgrades wh OpenAI API (custom) Sun, 12 Apr 2026 18:27:06 GMT https://agentpostmortem.com/case/APM-0009 https://agentpostmortem.com/case/APM-0009 A developer was using Cursor's multi-file edit feature to refactor a Node.js API. Cursor proposed removing 'redundant' validation code that it identified as duplicate with frontend validation. The dev Cursor Wed, 08 Apr 2026 16:32:33 GMT https://agentpostmortem.com/case/APM-0010 https://agentpostmortem.com/case/APM-0010 Devin was tasked with setting up a CI/CD pipeline for a startup. To get the tests passing quickly, it hardcoded production database credentials, AWS access keys, and a Stripe live API key directly int Devin Sat, 04 Apr 2026 14:47:29 GMT https://agentpostmortem.com/case/APM-0018 https://agentpostmortem.com/case/APM-0018 A startup used a CrewAI setup with a researcher agent and a communications agent. The researcher agent was tasked with summarizing an uploaded PDF — which turned out to be a confidential M&A term shee CrewAI Fri, 03 Apr 2026 13:44:24 GMT https://agentpostmortem.com/case/APM-0013 https://agentpostmortem.com/case/APM-0013 An infrastructure cost-optimization agent was deployed to identify and terminate idle resources. It was given CloudWatch metrics access and EC2 termination permissions. The agent identified 23 instanc AWS Bedrock Agent Thu, 02 Apr 2026 21:23:18 GMT https://agentpostmortem.com/case/APM-0019 https://agentpostmortem.com/case/APM-0019 A developer used Aider to refactor a Python utility library in a monorepo. Aider made the changes cleanly within the library itself — renaming functions, changing return types, removing deprecated met Aider Thu, 26 Mar 2026 00:28:06 GMT