Skip to content
Back to search

ventry

MarketplaceWorks with Claude CodeGitHub

ventry: an inventory and usage audit of your Claude Code setup.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/sedy2014--claude-ventry/badge.svg)](/listings/sedy2014--claude-ventry)

HTML

<a href="/listings/sedy2014--claude-ventry"><img src="/api/listings/sedy2014--claude-ventry/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

/plugin marketplace add sedy2014/claude-ventry
/plugin install ventry@ventry

Local testing before publishing: `/plugin marketplace add C:\repos\ventry`

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 22 lines
  1. {
  2. "$schema": "https://json.schemastore.org/claude-code-marketplace.json",
  3. "name": "ventry",
  4. "description": "ventry: an inventory and usage audit of your Claude Code setup.",
  5. "owner": {
  6. "name": "Sidharth Mahotra",
  7. "url": "https://www.linkedin.com/in/sidharthmahotra"
  8. },
  9. "plugins": [
  10. {
  11. "name": "ventry",
  12. "source": "./",
  13. "description": "Inventory of your Claude Code hooks, plugins, MCP servers, skills, slash commands and subagents, exported as Obsidian-friendly markdown with usage counts from local transcripts.",
  14. "category": "productivity",
  15. "tags": ["inventory", "audit", "obsidian", "usage"],
  16. "keywords": ["inventory", "audit", "hooks", "plugins", "mcp", "skills", "obsidian", "usage"],
  17. "homepage": "https://github.com/sedy2014/claude-ventry",
  18. "repository": "https://github.com/sedy2014/claude-ventry",
  19. "license": "MIT"
  20. }
  21. ]
  22. }
README.md · README · 71 lines
  1. # ventry: Claude Code inventory
  2. Know what your Claude Code is running. ventry inventories your **hooks, plugins, MCP servers, skills, slash commands and subagents** and writes them as Obsidian-friendly markdown notes, with **real usage counts** mined from your local session transcripts, so you can see what's installed but never used.
  3. > Status: v0.1 (inventory export). Session-start suggestions (status line) are planned, see Roadmap.
  4. ## What you get
  5. A folder of notes (default `~/.claude/ventry/`, or any folder, e.g. inside an Obsidian vault):
  6. | Note | Contents |
  7. |---|---|
  8. | `claude-code-inventory.md` | counts per area + **installed but never used** list |
  9. | `plugins.md` | plugin, marketplace, version, scope, enabled |
  10. | `hooks.md` | source, event, matcher, command (across user/project/local/plugin) |
  11. | `mcp-servers.md` | server, source, transport, calls, last used |
  12. | `skills.md`, `commands.md`, `agents.md` | description, calls, last used |
  13. Every note has YAML frontmatter (`type`, `generated`, `count`, `tags`) so Dataview queries work, and `claude-code-inventory.md` links the rest with `[[wikilinks]]`.
  14. ## Install
  15. ```
  16. /plugin marketplace add sedy2014/claude-ventry
  17. /plugin install ventry@ventry
  18. ```
  19. Local testing before publishing: `/plugin marketplace add C:\repos\ventry`
  20. ## Use
  21. - **Automatic:** a `SessionStart` hook refreshes the notes in the background (throttled to once per 6 hours, prints nothing, never delays startup).
  22. - **On demand:** `/ventry:ventry` (optionally `--out DIR`, `--no-usage`) runs the export and summarises prune candidates.
  23. - **Standalone:** `python scripts/ventry.py --out ~/Vault/claude-ventry`
  24. Set `VENTRY_OUT` to change the default output folder.
  25. ### Per-project snapshots
  26. Project settings (a project's hooks, `.mcp.json`, skills and commands) differ from project to project, but the main notes show only the project that ran last. Add `--per-project` to also keep a snapshot for each project:
  27. ```text
  28. ~/.claude/ventry/ latest overall view
  29. ~/.claude/ventry/projects/<name>-<id>/ one folder per project
  30. ```
  31. `<id>` is a short hash of the project's full path, so two projects with the same folder name stay apart. Links inside a snapshot point to that snapshot's own notes. The SessionStart hook uses `--per-project`, with its own 6-hour timer per project, so a new project gets its snapshot at its first session. On demand: `/ventry:ventry --per-project`.
  32. ## Privacy
  33. ventry only reads local files (`~/.claude/settings.json`, project `.claude/`, plugin folders, `~/.claude/projects/*.jsonl`). It extracts tool *names* and slash-command names only, never prompt or code content, and it makes **no network calls**.
  34. Besides the notes, it writes two small files to the output folder (`<claude dir>/ventry/` by default, or wherever `--out` points): `.last-run` (the throttle timestamp) and `.ventry-cache.json` (per-transcript counts).
  35. The notes themselves contain your project's absolute path, full hook commands, and MCP server URLs or command lines. Those can include tokens or other secrets, so keep the output folder private and do not sync it to a shared vault.
  36. ## Limits (be aware)
  37. - Hook firings are not recorded in transcripts, so hooks have no counts. Use `claude --debug`.
  38. - "Unused" means unused in the transcripts still on disk. Treat it as a prune *candidate*.
  39. - Settings, plugin registry and transcript formats can change between Claude Code versions; ventry parses defensively and fails quietly.
  40. - Requires Python 3 on PATH (`python3` or `python`). The script uses only the standard library. The unit tests need Python 3.10 or newer.
  41. ## Roadmap
  42. 1. Session-start suggestions via the status line, driven by an editable `rules/suggestions.yaml` (repo signals to suggested MCP servers / hooks). Suggest only, once per item per session, with accept/dismiss logging.
  43. 2. Optional per-item notes for richer Dataview queries.
  44. ## License
  45. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.