DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to replay exactly what your coding agent did with Agent Beacon

Open one trace per session with prompts, tool calls, commands, edits, approvals, MCP activity and tokens.

Short answer: Install Beacon and run beacon endpoint install, choosing Local; then run your agent as usual in Claude Code, Codex, Cursor or another supported harness; run beacon traces to browse traces, event timelines, token usage and retained content in the terminal; nothing is sent anywhere. All steps.

Credentials. Traces retain prompt and file content, so anyone who can read ~/.beacon or install a lens can see it; the README warns to install only lenses you trust with your traces.

Who it is forA developer whose agent changed files overnight and who wants to see every command and approval before merging.

The problemAgent sessions vanish when they end, and each tool keeps its own partial, proprietary history.

The outcomeA local, read-only replay of each session across harnesses, plus the raw JSONL to search yourself.

What you need first

  • macOS with Homebrew, Linux (Debian, Ubuntu, Fedora, RHEL, Rocky or Alma, as .deb or .rpm) or Windows x64 (MSI)
  • At least one supported harness, such as Claude Code, Codex CLI, Cursor or OpenCode
  • A beacon.sh sign-in only if you choose Beacon Cloud; Local needs no account
  • For forwarding: an existing destination such as Splunk HEC, Datadog, Elastic, Microsoft Sentinel or an S3 or GCS bucket

Steps

  1. Install Beacon and run beacon endpoint install, choosing Local
  2. Run your agent as usual in Claude Code, Codex, Cursor or another supported harness
  3. Run beacon traces to browse traces, event timelines, token usage and retained content in the terminal; nothing is sent anywhere
  4. Run beacon endpoint dashboard for the local web view and open a session's lenses, such as files changed with diffs or an itemized token bill
  5. Read or grep the raw event stream in ~/.beacon/endpoint/logs/runtime.jsonl

Prompt for your agent

Using Asymptote-Labs/agent-beacon on this machine, open the trace for my last <harness> session with `beacon traces` and list every command it ran, every file it edited and every approval, with total tokens. Verify the counts against ~/.beacon/endpoint/logs/runtime.jsonl.

Caveat. Coverage differs by harness: for example, Codex CLI does not report file activity and Cursor does not report tokens in the support table.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for Asymptote-Labs/agent-beacon and "Replay exactly what your coding agent did"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer