{
  "id": 18816757,
  "title": "Your AI coding agent has amnesia — here's a 30-second fix",
  "url": "https://urgent.news/2026/11/07/your-ai-coding-agent-has-amnesia-heres-a-30-second-fix",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-07T23:51:29.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/r3v3rs3/your-ai-coding-agent-has-amnesia-heres-a-30-second-fix-3jpa"
  },
  "original_language": "en",
  "account": "Every day, an AI coding agent starts with no knowledge of you. Yesterday, you informed it about the project using PostgreSQL 16, payments via Stripe with idempotent webhooks, the client's disapproval of em-dashes, and that the legacy folder is off-limits. However, today, the agent has forgotten all this. Re-explaining is not a prompt issue, but a memory problem. Corvusmem provides the coding agent (Claude Code, Codex, or a simple CLI) with a memory that remains persistent and doesn't leave your machine. The memory is stored as a simple file in your repository, without any server, cloud, API key, telemetry, embeddings, or vector databases. It's deterministic, same inputs produce the same results, and has zero dependencies. The memory is language-agnostic and can be shared among team members, with each person's facts staying separate and merging cleanly. To install Corvusmem, run `npx corvusmem install`, then save information using `npx corvusmem --save --dimension project.stack --value \"API is Fastify v5 on Node 22, PostgreSQL 16 + pgvector\"` with keys and values like \"stack\" and \"backend\". To recall information, simply run `npx corvusmem --recall backend`. The agent will capture and recall memory on its own, eliminating the need for constant babysitting. There's a limit on how much memory gets injected per message to keep the context window small. You can visualize your memory as a local HTML brain by running `npx corvusmem --cortex`, showing decisions, goals, project facts, and more in their own zones. Remember, Corvusmem is a new tool and the author is not exaggerating download counts. It is open-source for non-commercial use with a separate commercial license. If you find any issues, give it a try and report them. For more details and installation instructions, visit https://corvusmem.vercel.app/.",
  "summary": "Every morning your AI coding agent wakes up with no idea who you are. You told it yesterday that the project is on PostgreSQL 16, that payments go through Stripe with idempotent webhooks, that the client hates em-dashes, that the legacy/ folder is off-limits. Today? Blank slate. You re-explain it. Tomorrow you re-explain it again. That's not a prompt problem. It's a memory problem. Corvusmem…",
  "key_points": [
    "Corvusmem provides persistent memory for AI coding agents.",
    "Memory stored as file in repository, no cloud or API keys.",
    "Install via npx corvusmem install and save with --save."
  ],
  "editors_take": "Localizing memory within the repository eliminates the repetitive burden of re-explaining project context to AI agents while ensuring data privacy through a serverless, deterministic architecture.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}