<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Ming </title>
    <description>The latest articles on DEV Community by Ming  (@maxont).</description>
    <link>https://dev.to/maxont</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4247250%2F6b8a1884-cdce-45ba-89b7-3540ab4a9d88.jpg</url>
      <title>DEV Community: Ming </title>
      <link>https://dev.to/maxont</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/maxont"/>
    <language>en</language>
    <item>
      <title>existence-trace: a journal your AI reads, and has to prove it used</title>
      <dc:creator>Ming </dc:creator>
      <pubDate>Mon, 16 Nov 2026 08:54:55 +0000</pubDate>
      <link>https://dev.to/maxont/existence-trace-a-journal-your-ai-reads-and-has-to-prove-it-used-3p8i</link>
      <guid>https://dev.to/maxont/existence-trace-a-journal-your-ai-reads-and-has-to-prove-it-used-3p8i</guid>
      <description>&lt;p&gt;existence-trace is an open-source CLI (&lt;code&gt;etrace&lt;/code&gt;) for an AI that reads your journal every day, asks you where it is unsure, turns your answers into rules, and has to prove it used them. It runs on your own files with any model, under the MIT license.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;You spend hours a week with Claude, ChatGPT or Codex. They learn your stack, your repo and your naming style. What they pick up about you ends up in a vendor's memory you cannot fully check, or nowhere at all. existence-trace is the missing half: a journal your AI actually reads, in files you own.&lt;/p&gt;

&lt;h2&gt;
  
  
  How it works
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;You write.&lt;/strong&gt; A few lines a day, or import old diary files (&lt;code&gt;.md&lt;/code&gt;, &lt;code&gt;.txt&lt;/code&gt;, Day One &lt;code&gt;.json&lt;/code&gt;) with one command.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;It reads.&lt;/strong&gt; Each day it writes a report through three lenses: emotion (PAD), stability and exploration (G), growth (XP).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;It asks.&lt;/strong&gt; When it is unsure about you, it asks one question with evidence instead of guessing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;It proves.&lt;/strong&gt; Your answer becomes a rule. When a rule changes a report, it records a receipt with the before and the after. A weekly audit fails if a rule was never applied.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Your own words stay word for word in a protected block that nothing can rewrite; the AI only writes its own blocks. PAD, G and XP are reflection aids, not clinical scales.&lt;/p&gt;

&lt;h2&gt;
  
  
  Compared with built-in chat memory
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;/th&gt;
&lt;th&gt;Built-in chat memory&lt;/th&gt;
&lt;th&gt;existence-trace&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Where it lives&lt;/td&gt;
&lt;td&gt;the vendor's account&lt;/td&gt;
&lt;td&gt;plain text files in your own folder, versioned with git&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Who decides what is remembered&lt;/td&gt;
&lt;td&gt;mostly the model&lt;/td&gt;
&lt;td&gt;you: it asks, you answer, the answer becomes a versioned rule&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Proof it used what it knows&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;td&gt;a receipt with before and after, and a weekly audit&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Which AI&lt;/td&gt;
&lt;td&gt;one vendor&lt;/td&gt;
&lt;td&gt;Anthropic, OpenAI, DeepSeek, OpenRouter, Ollama, or any OpenAI-compatible address&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;It needs Python 3.11+ and uses the standard library only.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -fsSL https://maxont.github.io/existence-trace/install.sh &lt;span class="o"&gt;|&lt;/span&gt; bash
etrace ai &lt;span class="nb"&gt;set&lt;/span&gt;                            &lt;span class="c"&gt;# connect an AI&lt;/span&gt;
etrace import ~/Documents/diary --run    &lt;span class="c"&gt;# bring in your old diary and let the AI read it&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Add &lt;code&gt;--dry-run&lt;/code&gt; to the installer or to &lt;code&gt;etrace import&lt;/code&gt; to preview first. With no old diary, run &lt;code&gt;etrace init&lt;/code&gt;, then &lt;code&gt;etrace note&lt;/code&gt; to open today's page and &lt;code&gt;etrace run daily&lt;/code&gt; to read it. Your AI key is kept outside your data folder.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where it fits
&lt;/h2&gt;

&lt;p&gt;existence-trace is about one person's own record. Its sibling projects are about agents. &lt;a href="https://github.com/MaxonT/aisync" rel="noopener noreferrer"&gt;aisync&lt;/a&gt; handles Multi-Agent Context Management, Coordination &amp;amp; Observability across AI coding tools, and &lt;a href="https://github.com/MaxonT/TRACES.md" rel="noopener noreferrer"&gt;TRACES.md&lt;/a&gt; is a format for recording what each agent did. They are separate repos; existence-trace does not depend on either.&lt;/p&gt;

&lt;h2&gt;
  
  
  Limits
&lt;/h2&gt;

&lt;p&gt;This is a young project and the numbers it reports are reflection aids, not measurements. It is not a clinical or therapy tool. It sends your journal text to whichever AI service you connect, so choose one you are comfortable with, or use a local model through Ollama.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Repo (MIT): &lt;a href="https://github.com/MaxonT/existence-trace" rel="noopener noreferrer"&gt;https://github.com/MaxonT/existence-trace&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Site: &lt;a href="https://maxont.github.io/existence-trace/" rel="noopener noreferrer"&gt;https://maxont.github.io/existence-trace/&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>ai</category>
      <category>opensource</category>
      <category>python</category>
      <category>productivity</category>
    </item>
    <item>
      <title>TRACES.md: a plain-text flight recorder for what AI agents did</title>
      <dc:creator>Ming </dc:creator>
      <pubDate>Mon, 16 Nov 2026 08:43:39 +0000</pubDate>
      <link>https://dev.to/maxont/tracesmd-a-plain-text-flight-recorder-for-what-ai-agents-did-1nom</link>
      <guid>https://dev.to/maxont/tracesmd-a-plain-text-flight-recorder-for-what-ai-agents-did-1nom</guid>
      <description>&lt;p&gt;TRACES.md is an open format for recording what AI agents did. AGENTS.md tells an agent how to behave. TRACES.md is the file where it writes down what it actually did.&lt;/p&gt;

&lt;p&gt;I run several coding agents (Cursor, Trae, Kiro, Claude Code, Codex) on the same repositories. When something broke, I had no plain record of which agent ran which command and how it ended. TRACES.md came out of that.&lt;/p&gt;

&lt;h2&gt;
  
  
  The format
&lt;/h2&gt;

&lt;p&gt;A trace is one plain, append-only file per task. One line per operation: when, which tool, what it did, how it ended. It records no reasoning and no story, and it records failures.&lt;/p&gt;

&lt;p&gt;A minimal file, &lt;code&gt;.trace/2026-10-12-claude-code-fix-date-parser.md&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;task: Fix the failing date parser test
agent: claude-code
started: 2026-10-12T09:14:02-05:00
spec: traces/0.1.0

2026-10-12T09:14:31-05:00 | shell: python -m pytest tests/test_dates.py | run the date tests | fail: exit 1, 1 failed, 11 passed
2026-10-12T09:15:02-05:00 | edit: src/dates.py | change the parse format string to %Y-%m-%d | ok: 1 line changed
2026-10-12T09:15:20-05:00 | shell: python -m pytest tests/test_dates.py | run the date tests | ok: exit 0, 12 passed
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Each step is &lt;code&gt;ts | tool | action | result&lt;/code&gt;, and the result is &lt;code&gt;ok&lt;/code&gt;, &lt;code&gt;fail&lt;/code&gt; or &lt;code&gt;partial&lt;/code&gt; plus a factual detail. The same trace also exists as JSONL with a JSON Schema, and &lt;code&gt;traces fmt&lt;/code&gt; converts between the two without loss.&lt;/p&gt;

&lt;h2&gt;
  
  
  The CLI
&lt;/h2&gt;

&lt;p&gt;It needs Python 3.9+ and the standard library only.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;traces init --agent claude-code --task &lt;span class="s2"&gt;"fix the date parser"&lt;/span&gt;
&lt;span class="kd"&gt;export&lt;/span&gt; &lt;span class="nv"&gt;TRACES_FILE&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;.trace/&lt;span class="o"&gt;&amp;lt;&lt;/span&gt;printed file&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;
traces run -- python -m pytest
traces append -t &lt;span class="s2"&gt;"edit: src/dates.py"&lt;/span&gt; -a &lt;span class="s2"&gt;"fix the format string"&lt;/span&gt; -s ok -d &lt;span class="s2"&gt;"1 line changed"&lt;/span&gt;
traces validate
traces stats
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;traces run&lt;/code&gt; runs a command and records its real exit code. &lt;code&gt;traces validate&lt;/code&gt; exits 1 if a trace is invalid or hides a failure, for example an &lt;code&gt;ok&lt;/code&gt; next to a non-zero exit code, or a retry whose failure was never recorded. &lt;code&gt;traces stats&lt;/code&gt; reports per agent, per tool, and the failure rate.&lt;/p&gt;

&lt;h2&gt;
  
  
  One pitfall if you record from hooks
&lt;/h2&gt;

&lt;p&gt;Many hook systems fire their "after tool" event only when a call succeeds. In Claude Code, &lt;code&gt;PostToolUse&lt;/code&gt; never sees a failed call. That arrives as &lt;code&gt;PostToolUseFailure&lt;/code&gt;. A hook on the success event alone writes a trace that looks clean while every failure is missing. Register the failure event too, then make a command fail on purpose and check that a &lt;code&gt;fail&lt;/code&gt; step shows up.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where it fits
&lt;/h2&gt;

&lt;p&gt;TRACES.md is the observability piece of a larger setup. The sibling project is &lt;a href="https://github.com/MaxonT/aisync" rel="noopener noreferrer"&gt;aisync&lt;/a&gt;: Multi-Agent Context Management, Coordination &amp;amp; Observability. aisync keeps shared rules and a journal across every AI coding tool, and coordinates who is editing what. TRACES.md is the per-task record of what each agent did. aisync is a separate repo and tool; TRACES.md works without it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Limits
&lt;/h2&gt;

&lt;p&gt;This is version 0.1 of the spec, so expect changes. The repo is small and has no adoption numbers I would put in an article. It is not affiliated with AGENTS.md. It borrows that project's shape: one plain Markdown convention any agent can follow with no SDK.&lt;/p&gt;

&lt;p&gt;The repository writes its own traces in &lt;code&gt;.trace/&lt;/code&gt;, failures included.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Repo (MIT): &lt;a href="https://github.com/MaxonT/TRACES.md" rel="noopener noreferrer"&gt;https://github.com/MaxonT/TRACES.md&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Spec: &lt;a href="https://github.com/MaxonT/TRACES.md/blob/main/SPEC.md" rel="noopener noreferrer"&gt;https://github.com/MaxonT/TRACES.md/blob/main/SPEC.md&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Site: &lt;a href="https://maxont.github.io/TRACES.md/" rel="noopener noreferrer"&gt;https://maxont.github.io/TRACES.md/&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>ai</category>
      <category>opensource</category>
      <category>python</category>
      <category>productivity</category>
    </item>
    <item>
      <title>aisync: one local hub for rules and coordination across AI coding agents</title>
      <dc:creator>Ming </dc:creator>
      <pubDate>Mon, 16 Nov 2026 08:24:38 +0000</pubDate>
      <link>https://dev.to/maxont/aisync-one-local-hub-for-rules-and-coordination-across-ai-coding-agents-4egp</link>
      <guid>https://dev.to/maxont/aisync-one-local-hub-for-rules-and-coordination-across-ai-coding-agents-4egp</guid>
      <description>&lt;p&gt;aisync is a local hub for Multi-Agent Context Management, Coordination &amp;amp; Observability.&lt;/p&gt;

&lt;p&gt;If you run Claude Code, Codex and Cursor on the same machine, each tool keeps its own rules, two of them can edit the same file, and nothing records who did what. aisync is a small local hub that tries to fix those three things.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it is
&lt;/h2&gt;

&lt;p&gt;aisync keeps everything in &lt;code&gt;~/.ai-sync&lt;/code&gt;, which is a plain git repo. It is agent-agnostic at its core. It ships native hooks for Claude Code, Codex and Cursor, and other agents can use the same hub through a CLI, an MCP server or shared instruction files.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it does
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Shared rules.&lt;/strong&gt; Rules are shared across tools and synced with a 3-way merge, so a rule written once reaches every agent.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A verifiable journal.&lt;/strong&gt; Agent activity is written to a hash-chained journal. &lt;code&gt;aisync verify&lt;/code&gt; checks the chain for edits.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;File claims.&lt;/strong&gt; An agent declares the files it is about to touch. An overlapping claim is detected and reported with exit code 3, so two agents can avoid editing the same file.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Mid-session news.&lt;/strong&gt; Where hooks are available, an agent is told what other agents did while it was working.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Limits
&lt;/h2&gt;

&lt;p&gt;File claims are cooperative, not OS-level locks: agents have to use the claim protocol. Hash-chain verification detects changes to recorded events. It does not prove that every claimed action happened.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;It is MIT licensed, needs Python 3.11 or newer and has no runtime dependencies.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;pip install aisync-cli
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Source and docs: &lt;a href="https://github.com/MaxonT/aisync" rel="noopener noreferrer"&gt;https://github.com/MaxonT/aisync&lt;/a&gt;&lt;br&gt;
Website: &lt;a href="https://maxont.github.io/aisync/" rel="noopener noreferrer"&gt;https://maxont.github.io/aisync/&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>opensource</category>
      <category>python</category>
      <category>productivity</category>
    </item>
  </channel>
</rss>
