My Claude Code Mods plugin
A marketplace of 17 Claude Code mods: token-spend breakdown, a rollback edit timeline, assumption tracking, a verification board and a live command feed
- 1,099 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Time Machine keeps shadow snapshots of the project's files through git and can restore their content to a past step
- Agent Profiler, once configured, sends session traces to an external OTLP collector
- The mods' hooks intercept and log every Claude Code tool call in a session
Install
- Works in
Manual install
claude plugin marketplace add YeonwooSung/my-claude-code-modsAfter adding the marketplace, install mods one at a time: claude plugin install time-machine@my-claude-code-mods --scope user.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Time Machine keeps shadow snapshots of the project's files through git and can restore their content to a past step
- Agent Profiler, once configured, sends session traces to an external OTLP collector
- The mods' hooks intercept and log every Claude Code tool call in a session
This is third-party code. Review the repository files before installing.
What it does
My Claude Code Mods is the author's personal catalog, set up as a Claude Code plugin marketplace through `.claude-plugin/marketplace.json`: one `claude plugin marketplace add` command connects all 17 mods at once, and each is then installed separately with `claude plugin install <name>@my-claude-code-mods`. Time Machine records every agent tool call as a file-state snapshot through a shadow Git repository: you can open any past step, see the files as they were at that point, find which edit broke a test, restore it, branch the history and re-run a step with the same input. Agent Profiler shows where a session's time goes (model wait, tools, subagents, re-reads of the same file) and optionally exports OTLP traces for an external Jaeger or Tempo. Loop Detective watches for repeated Bash failures and edit/retry cycles, gathers evidence and surfaces investigation prompts in its own live session pane. Seven newer mods (verify-board, scope-guard, blast-radius, decision-ledger, handoff-pack, command-shelf, env-doctor, dependency-lens, assumption-tracker, review-queue) still work as plain text commands with no live panel yet, and are installed from a local checkout with the `--plugin-dir` flag. Verify-board keeps verification results tied to a code revision with pass, fail, stale and unknown statuses, decision-ledger records explicit decisions with their rationale and rejected alternatives, and dependency-lens shows direct and transitive npm dependency changes along with install-script markers.
Who it is for. For developers who work with Claude Code every day and want to see where a session's time and tokens go, roll back to earlier file versions, and keep an explicit log of the agent's decisions and assumptions.
Good fit when
- The agent broke several files and a plain git diff isn't enough: you need to see exactly which tool call led to the current state
- A session got expensive or slow and you need to see where the time goes — model wait, tools, or repeated file reads
- A Bash command keeps failing in a loop and you need gathered evidence for the cause instead of rereading the whole log by hand
- You want an explicit decision log with rejected alternatives you can show a colleague or revisit a week later
Not a fit when
- You need Claude Code older than 2.1.287: the mods only work from the version where mod support is enabled
- You want Time Machine without git on the machine: its shadow snapshots are built through git, and it won't start without it
- You want a live panel for one of the seven newer mods: they currently respond only with command text, with no interface panel yet
- You want a mod that fixes the problem it finds: all seventeen mods only show data and logs, none of them edit your code
Example request
Install time-machine and agent-profiler from my-claude-code-mods, I want to see the edit history and where session time goesLimitations
License is Apache-2.0.
- Requires a Claude Code build with mod support, enabled by default from version 2.1.287; Token Run was specifically verified on 2.1.296 and 2.1.301.
- Time Machine additionally needs git installed: without it, its shadow file snapshots are not created.
- Seven mods from the second wave (verify-board, scope-guard, blast-radius, decision-ledger, handoff-pack, command-shelf, env-doctor, dependency-lens, assumption-tracker, review-queue) are preview-only from a local checkout via `claude --plugin-dir ./<mod-name>` and are not yet in the shared mod marketplace.
- Agent Profiler sends traces to an external OTLP collector only if you configure one yourself; without that setup the data stays local.
- Installing all 17 mods with one command stops at the first failure: if one mod fails to install, the ones listed after it are not installed automatically.
How to disable. Run `claude plugin uninstall <name>@my-claude-code-mods` for a specific mod, or open `/plugin` in a Claude Code session and disable the mods you want from the loaded list.
FAQ
Can I install just one mod instead of all 17?
Yes, after one claude plugin marketplace add command, claude plugin install <name>@my-claude-code-mods for a single mod is enough. The author's one-shot install-everything block is optional and can be skipped.
What sets Time Machine apart from a plain git diff?
A git diff only shows the final file difference. Time Machine keeps the order of tool calls and a file snapshot after each one, so you can open the state at step ③ separately from step ⑤ and see which edit caused the test failure.
Where do Agent Profiler's traces go?
By default, session timing data stays local. OTLP export for Jaeger or Tempo only turns on if you configure an external collector yourself; without that, nothing leaves the machine.
Why don't the seven newer mods have a live panel?
The author labels them as first releases that respond only with command text; the live panels already present in Time Machine and Loop Detective haven't been built for them yet.
Do I need Node.js or a repository clone for the normal install?
No, installing through claude plugin marketplace add and claude plugin install needs no clone and no npm install. Cloning is only needed to preview the seven newer mods via --plugin-dir.
Discussion
No questions or reviews yet.
Related
PluginsA marketplace of fourteen plugins for Claude Code power users: model council, email, calendar, video digests and routine tasks
A Claude Code plugin that catches your corrections in chat and, after review, writes them into CLAUDE.md and AGENTS.md
A Claude Code plugin: a status line with context, 5-hour and 7-day limits, cost and usage of Codex, Gemini and z.ai
daymade Skills Marketplace
Claude Code Skills Marketplace
A plugin marketplace for Claude Code and Codex: documents, audio, GitHub, session history, finance, macOS and a hardened skill-creator
A statusline for Claude Code: token and cost usage, context left, rate limits and git branch in four styles and six themes
A business operating system for Claude Code: a morning digest across mail, Slack, payments, analytics. The wizard installs a background daemon and programs
README in short
The repository description is a list of seventeen mods, one line each with a link to its own README, several duplicated in Korean. A block on previewing the newer mods locally with --plugin-dir follows, plus detailed install instructions: one command for all of them or one at a time, from a terminal or slash commands inside a session. A short 'Development' section at the end links to internal material for anyone who wants to add their own mod.