toolmenu 0.25.2 → 0.27.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (60) hide show
  1. package/README.md +170 -683
  2. package/dist/args.d.ts +14 -2
  3. package/dist/args.js +44 -1
  4. package/dist/auth-check.d.ts +27 -0
  5. package/dist/auth-check.js +177 -0
  6. package/dist/auth.d.ts +7 -0
  7. package/dist/auth.js +29 -1
  8. package/dist/auto.d.ts +8 -0
  9. package/dist/auto.js +9 -2
  10. package/dist/breakdown.js +14 -10
  11. package/dist/cli.js +151 -49
  12. package/dist/clients.d.ts +13 -2
  13. package/dist/clients.js +65 -11
  14. package/dist/config.js +6 -0
  15. package/dist/diff.d.ts +35 -1
  16. package/dist/diff.js +57 -12
  17. package/dist/failures.js +4 -3
  18. package/dist/help.d.ts +1 -1
  19. package/dist/help.js +45 -18
  20. package/dist/history.js +5 -0
  21. package/dist/html.d.ts +29 -0
  22. package/dist/html.js +360 -0
  23. package/dist/index.d.ts +3 -2
  24. package/dist/index.js +3 -2
  25. package/dist/init.d.ts +14 -0
  26. package/dist/init.js +20 -6
  27. package/dist/junit.d.ts +6 -1
  28. package/dist/junit.js +12 -3
  29. package/dist/mcp-env.d.ts +6 -0
  30. package/dist/mcp-env.js +6 -0
  31. package/dist/mcp.d.ts +11 -1
  32. package/dist/mcp.js +123 -22
  33. package/dist/menu.d.ts +6 -0
  34. package/dist/menu.js +12 -1
  35. package/dist/options.d.ts +10 -0
  36. package/dist/options.js +3 -0
  37. package/dist/report.d.ts +14 -0
  38. package/dist/report.js +57 -51
  39. package/dist/review.d.ts +48 -0
  40. package/dist/review.js +212 -0
  41. package/dist/rule-docs.json +49 -4
  42. package/dist/rules/clients.d.ts +11 -0
  43. package/dist/rules/clients.js +65 -2
  44. package/dist/rules/determinism.d.ts +3 -0
  45. package/dist/rules/determinism.js +44 -7
  46. package/dist/rules/ids.js +5 -1
  47. package/dist/rules/index.js +5 -2
  48. package/dist/rules/output.d.ts +21 -0
  49. package/dist/rules/output.js +49 -0
  50. package/dist/rules/search.js +1 -1
  51. package/dist/rules/spec.d.ts +19 -0
  52. package/dist/rules/spec.js +86 -1
  53. package/dist/sanitize.d.ts +42 -0
  54. package/dist/sanitize.js +142 -0
  55. package/dist/scan.d.ts +9 -2
  56. package/dist/scan.js +46 -24
  57. package/dist/session.d.ts +17 -0
  58. package/dist/session.js +119 -13
  59. package/dist/snapshot.js +2 -1
  60. package/package.json +5 -3
package/README.md CHANGED
@@ -5,7 +5,7 @@
5
5
  <h1 align="center">toolmenu</h1>
6
6
 
7
7
  <p align="center">
8
- <b>Checks for MCP servers' tool lists: changes that confuse agents, break prompt caches or break clients, caught before your users hit them.</b>
8
+ <b>Tests your MCP server the way agents and clients meet it: the menu they read, the calls they make, the results and sign-in they get back, and every release after.</b>
9
9
  </p>
10
10
 
11
11
  <p align="center">
@@ -15,223 +15,108 @@
15
15
  [![license: MIT](https://img.shields.io/badge/license-MIT-blue)](https://github.com/niksa90/toolmenu/blob/main/LICENSE)
16
16
  [![MCP: 2026-07-28 and 2025](https://img.shields.io/badge/MCP-2026--07--28%20%C2%B7%202025-6E56CF)](https://modelcontextprotocol.io)
17
17
  [![toolmenu](https://img.shields.io/endpoint?url=https%3A%2F%2Fraw.githubusercontent.com%2Fniksa90%2Ftoolmenu%2Fmain%2Fdocs%2Ftoolmenu-badge.json)](https://github.com/niksa90/toolmenu)
18
- [![toolmenu MCP server on Glama](https://glama.ai/mcp/servers/niksa90/toolmenu/badges/score.svg)](https://glama.ai/mcp/servers/niksa90/toolmenu)
19
- [![toolmenu MCP on mcprush](https://mcprush.com/badge/toolmenu-mcp/score.svg)](https://mcprush.com/to/toolmenu-mcp?ref=badge)
20
- [![Listed on mcpservers.org](https://mcpservers.org/badge.svg)](https://mcpservers.org/servers/niksa90/toolmenu)
18
+ [![toolmenu MCP server on Glama](https://glama.ai/mcp/servers/niksa90/toolmenu/badges/score.svg)](https://glama.ai/mcp/servers/niksa90/toolmenu)
19
+ [![toolmenu MCP on mcprush](https://mcprush.com/badge/toolmenu-mcp/score.svg)](https://mcprush.com/to/toolmenu-mcp?ref=badge)
20
+ [![Listed on mcpservers.org](https://mcpservers.org/badge.svg)](https://mcpservers.org/servers/niksa90/toolmenu)
21
21
 
22
- </p>
22
+ ## License
23
23
 
24
- Your MCP server's tool list is the menu every agent reads before it does anything.
25
- toolmenu checks that menu the way clients and agents will meet it: across calls,
26
- connections, clients and releases. No LLM is involved, so the same inputs always give
27
- the same answer and it fits in CI.
24
+ MIT
28
25
 
29
- ```sh
30
- npx toolmenu snapshot -- node dist/server.js # or: npx toolmenu snapshot https://example.com/mcp
31
- npx toolmenu init -- node dist/server.js # CI: a baseline, a workflow, a comment on every PR
32
- ```
33
26
 
34
- ![toolmenu session output: tools inserted mid-list, a description edited, each pinned to the step that caused it](https://raw.githubusercontent.com/niksa90/toolmenu/main/docs/demo.svg)
27
+ </p>
35
28
 
36
- ## What it catches
29
+ An MCP server can pass its own tests and still fail in the client. The client's SDK
30
+ rejects a tool list the server's own test client accepted. A default built from the
31
+ clock makes every listing different, so no prompt cache ever matches. A patch release
32
+ makes a parameter required. None of it shows until someone's agent has fewer tools, or
33
+ a bigger bill, and no error to explain it.
37
34
 
38
- | | |
39
- |---|---|
40
- | **A menu that changes mid-session** | `session` calls your tools and watches the list. A tool added mid-conversation can rewrite the whole cached prompt: one unlock cost 40–57K tokens of cache in my logs. |
41
- | **Breaking releases** | `diff` compares two menus: removed tools, new required parameters, narrowed types, token growth. It also says whether your version bump is big enough. |
42
- | **The 2026-07-28 protocol** | The tool set must not vary per connection, and results need `resultType`. Elicitation, multi round-trip input and `server/discover` are checked too, as are the 2025 handshakes. |
43
- | **MCP Apps, Tasks, Server Cards** | It reads `ui://` resources as a host does and checks who may see each tool. It catches tasks sent to clients that can't poll them, and Server Cards that contradict the live server. |
44
- | **Your client's view** | `--client claude-code \| claude-api-search \| openai` shows what each client loads, cuts, can find and refuses. |
45
- | **What agents misread** | It flags vague `id`s, IDs no tool returns, instructions buried late in descriptions, and writes without annotations or marked read-only. |
46
- | **Words routing to the wrong tool** | `routes.yml` pins which tool each keyword must (and must not) reach, the way a keyword or BM25 tool search sees the menu. A description edit that lets `get_audit_trail` win "audit" fails the check, with no model and no eval budget. [More](#snapshot-rules) |
47
- | **Menus that drift** | It catches defaults built from the clock or a hash order, menus that differ per process, per connection or offline, and listings that depend on who asks. |
48
-
49
- Every finding says what's wrong, where it is and what toolmenu saw, and ends with a `→ Next:` step.
50
- It runs on the 22-server [bench](bench/README.md) and on hosted servers (GitHub, Atlassian, Sentry,
51
- Notion, Supabase, Vercel…). What it has found on real servers is in
52
- [docs/FINDINGS.md](https://github.com/niksa90/toolmenu/blob/main/docs/FINDINGS.md).
53
-
54
- > **Status: 0.25.** Spec in [docs/SPEC.md](https://github.com/niksa90/toolmenu/blob/main/docs/SPEC.md),
55
- > every rule in [docs/RULES.md](docs/RULES.md), changes in [CHANGELOG.md](CHANGELOG.md).
56
-
57
- <details>
58
- <summary><b>Why "when it changes" matters</b></summary>
59
-
60
- The MCP linters I found grade one snapshot of the menu: tool names, descriptions,
61
- schemas. toolmenu also checks **when** the menu changes:
62
-
63
- - **Mid-session.** Most clients put the tool list at the start of the prompt, so any
64
- change to it, even a tool added at the end, means everything after it is processed
65
- again. More on this, with a benchmark of the setups, in
66
- [this write-up](https://niksa.me/agent-tools-design-lessons). `session` pins each
67
- change to the step that caused it, and says where in the list it happened.
68
- - **Per connection.** 2026-07-28 says the tool set **must not** vary per connection or
69
- as a side effect of other requests (the 2025 versions allow it). Per-session
70
- "unlock" designs, like the one I built, are ruled out. `session` tells you whether
71
- yours is one of them. It also checks the menu clients will actually show: a broken
72
- `x-mcp-header` or a `$ref` they can't resolve hides a tool or fails every call to
73
- it. When a tool stops to ask the user something, `session` answers "no" and says
74
- whether the tool can take it.
75
- - **Between releases.** `diff` finds breaking changes and token growth, and checks the
76
- version bump. `history` does it across a package's published releases. It installs
77
- old versions with today's dependencies, so it records what they serve now
78
- (FINDINGS F4).
79
-
80
- Run against the reference server, `@modelcontextprotocol/server-everything`, it
81
- reports no false findings ([bench/](bench/README.md)).
82
-
83
- </details>
84
-
85
- ## Start here
86
-
87
- Needs Node 22 or later. The GitHub Action brings its own if the runner's Node is older,
88
- for toolmenu only: your server and the job's later steps keep the job's Node.
89
- Your server can be in any language: toolmenu starts it (stdio) or connects to it (HTTP).
90
-
91
- One command sets up CI: it snapshots your server into `menu.json` (the baseline) and
92
- writes `.github/workflows/toolmenu.yml` for your project (Node, Python, Go or Rust).
93
- Commit both, open a pull request, and toolmenu comments on it.
35
+ toolmenu does what a client does, in order, and checks each step against the spec, the
36
+ tool's own schemas and each client's documented limits:
94
37
 
95
- ```sh
96
- npx toolmenu init -- node dist/server.js # add --with-session to also run session --auto
97
- ```
38
+ ![What a client does with an MCP server, and what toolmenu checks at each step: sign-in, the connection, the menu, tool calls, the menu during a conversation, and the next release](https://raw.githubusercontent.com/niksa90/toolmenu/main/docs/overview.svg)
98
39
 
99
- Every value you pass with `--env` or `--header` goes into the workflow as
100
- `${{ secrets.… }}`, never as text (a name like `DATABASE_URL` doesn't say it holds a
101
- password); `init` says which secrets to add. It never overwrites a file.
102
-
103
- Run it anywhere in the repository. `.github/workflows/toolmenu.yml` goes to the git
104
- root, and the server starts in CI from the folder you ran `init` in, which is also
105
- where `menu.json` is saved. That's how a monorepo package works: run `init` in
106
- `packages/server`. Paths to files in the repository are made relative, so they work on
107
- the runner. `init` lists anything that won't work there (a file outside the
108
- repository, a `localhost` URL, a virtualenv the runner doesn't have), each with what
109
- to do.
110
-
111
- By hand:
40
+ No LLM is involved, so the same server gives the same findings on every run and it fits
41
+ in CI. It doesn't tell you whether an agent picks the right tool; that takes evals with a
42
+ model ([what it can't catch](#what-it-cant-catch)).
112
43
 
113
44
  ```sh
114
- npx toolmenu snapshot -- node dist/server.js # the menu, its findings, menu.json
115
- npx toolmenu diff menu.json new-menu.json # what changed between two releases
116
- npx toolmenu session --auto -- node dist/server.js # call read-only tools, watch the menu
45
+ npx toolmenu snapshot -- node dist/server.js # or: npx toolmenu snapshot https://example.com/mcp
46
+ npx toolmenu init -- node dist/server.js # CI: a baseline, a workflow, a comment on every PR
117
47
  ```
118
48
 
119
- HTTP servers work the same way: `npx toolmenu snapshot https://example.com/mcp --header "Authorization: Bearer $TOKEN"`.
120
- Servers behind OAuth (most hosted ones: GitHub, Atlassian, Sentry…): log in once, in a
121
- browser, and `snapshot` and `session` use that login from then on, refreshing it as
122
- needed:
49
+ Needs Node 22 or later. Your server can be in any language: toolmenu starts it (stdio)
50
+ or connects to it (HTTP).
123
51
 
124
- ```sh
125
- npx toolmenu auth login https://mcp.example.com/mcp
126
- npx toolmenu snapshot https://mcp.example.com/mcp
127
- ```
52
+ ![toolmenu session output: tools inserted mid-list, a description edited, each pinned to the step that caused it](https://raw.githubusercontent.com/niksa90/toolmenu/main/docs/demo.svg)
128
53
 
129
- Logins are stored per server in `~/.config/toolmenu/auth` (readable only by you);
130
- `auth list` and `auth logout <url>` manage them. A server that doesn't allow
131
- dynamic registration needs a pre-registered app: `--client-id` (and
132
- `--client-secret`). Over SSH or in a container, `auth login --no-browser` prints the
133
- login URL instead of opening it: open it on any device, sign in, and paste back the
134
- address the browser ends on (a `localhost` page that doesn't load is expected). In
135
- CI, where nobody can sign in, pass a token as a header.
136
- Both protocol generations are supported: 2026-07-28 (`server/discover`, stateless)
137
- and the 2025 `initialize` handshake, through the official TypeScript SDK.
138
-
139
- Every finding and error says what's wrong, where, what toolmenu saw, and ends in a
140
- `→ Next:` step. Findings marked `· unsure` are heuristics (names and words): they say
141
- what was seen, not what it means. `toolmenu explain <rule>` says what a rule checks,
142
- its default level and how to change it (every rule is in [docs/RULES.md](docs/RULES.md)).
143
- Every command has its own help: `toolmenu history --help`.
144
-
145
- Behind a proxy (`HTTPS_PROXY`), stdio servers get the proxy and CA settings
146
- (`HTTPS_PROXY`, `NO_PROXY`, `NODE_EXTRA_CA_CERTS`, `SSL_CERT_FILE`, `REQUESTS_CA_BUNDLE`…)
147
- along with the minimal environment, and `NODE_USE_ENV_PROXY=1` so a Node server's
148
- `fetch` uses the proxy. Without them a server that calls an API can't reach it, and
149
- fails as if it were broken.
150
-
151
- ### Every server you use: `scan`
54
+ **Contents:** [What it catches](#what-it-catches) · [A finding](#what-a-finding-looks-like) ·
55
+ [In CI](#in-ci-one-pr-comment) · [Commands](#commands) · [Tested on real servers](#tested-on-real-servers) ·
56
+ [What it can't catch](#what-it-cant-catch) · [Docs](#docs)
152
57
 
153
- ```sh
154
- npx toolmenu scan ~/Library/Application\ Support/Claude/claude_desktop_config.json
155
- npx toolmenu scan .mcp.json --format markdown
156
- ```
58
+ ## What it catches
157
59
 
158
- `scan` runs `snapshot` on every server in an MCP client config (Claude Desktop, Claude
159
- Code's `.mcp.json`, Cursor, VS Code: the `mcpServers` or `servers` object) or in a
160
- text file with one server per line (a URL or a command, optionally `name<TAB>` first),
161
- four at a time (`--concurrency`), and prints them side by side, then each server's
162
- findings, like this:
60
+ Each row is something a run on real servers found
61
+ ([docs/FINDINGS.md](https://github.com/niksa90/toolmenu/blob/main/docs/FINDINGS.md) has the versions and how to reproduce each).
163
62
 
164
- ```
165
- server tools tokens findings
166
- github 42 ~18K 1 error · 2 warn
167
- filesystem 11 ~2.1K no findings
168
- linear – – not checked: LINEAR_API_KEY is not set (the config refers to it)
169
- ```
63
+ | | What goes wrong | Seen on |
64
+ |---|---|---|
65
+ | **Lists and results a client rejects** | The official SDK client validates `tools/list` and every result. One bad schema and it drops the whole list; an answer that breaks the tool's own `outputSchema` becomes an error | 12 of 1,089 registry servers: 8 lose every tool, 4 lose answers |
66
+ | **A menu that never matches twice** | A default built from the clock, an order from a hash set, a container ID in a description. Every listing or every restart is a new menu, so cached prompts miss | PayPal (`list_transactions` dates), mcp-atlassian, Desktop Commander |
67
+ | **A menu that changes mid-conversation** | Most clients put tools first in the prompt, so a tool added by an "unlock" call means the conversation after it is processed again. 2026-07-28 rules per-connection menus out | One unlock re-processed 40–57K cached tokens in the author's own logs |
68
+ | **Releases that break callers** | Removed tools, new required parameters, narrowed types, a tool that became destructive, with the version bump the change needs | The official filesystem server made `move_file` destructive in a calendar release; Exa removed tools in patch releases |
69
+ | **Sign-in a strict client can't finish** | OAuth discovery that differs from the RFCs: a `resource` that isn't the server's URL, metadata that isn't there, PKCE not advertised | 12 of 79 vendor endpoints that answer 401 name a parent or slash-variant `resource` |
70
+ | **Tools one client can't use** | More tools than the client sends, names its API refuses or rewrites, descriptions cut before the instruction that matters | 6 servers in Docker's catalog list more than VS Code's 128 a request |
71
+ | **Tokens spent on nothing** | Unused `$defs`, the same schema block repeated across tools, growth per release | Notion's server: the same 9 unused `$defs` in all 24 tools, 72% of its menu |
170
72
 
171
- `${VAR}`, `${VAR:-default}` and `${env:VAR}` in the config are filled from the
172
- environment. A server whose variable isn't set isn't started with an empty
173
- credential: it's a row that says so. A server that couldn't be started or read
174
- makes the exit code 2, because it wasn't checked; one marked `disabled` is listed and
175
- left alone. URLs are printed without their query, which can hold a key. Nothing is
176
- written unless `--out-dir` names a folder for each server's menu file.
73
+ It also has heuristics for what agents misread (a parameter called just `id`, writes
74
+ with no annotations, two tools fighting over one word). Those are marked as guesses and
75
+ are never errors. Every rule is in [docs/RULES.md](https://github.com/niksa90/toolmenu/blob/main/docs/RULES.md).
177
76
 
178
- ## For the agent building your server: `mcp`
77
+ ## What a finding looks like
179
78
 
180
- `toolmenu mcp` is toolmenu as an MCP server (stdio), so Claude Code, Cursor or any
181
- other client can check the server it's working on while it works. In Claude Code, as a
182
- plugin (the server, and skills for checking a change, checking a release, and setting up CI):
79
+ A server whose `unlock_toolset` tool adds two tools to the menu mid-conversation, run
80
+ with `toolmenu session`:
183
81
 
184
82
  ```
185
- /plugin marketplace add niksa90/toolmenu
186
- /plugin install toolmenu@toolmenu
83
+ step 3: call unlock_toolset {"toolset":"audits"} · answered · menu changed · +2 tools · list_changed received
84
+ ERROR session/mid-insert
85
+ +2 tools inserted at position 1 (list_team_audits, get_team_audit). Invalidates the cached prompt:
86
+ the tool list and the conversation after it are processed again.
87
+ cost: the whole tool list (this server's part: ~406 tokens, estimate) and the conversation after it
88
+ → Next: Add new tools at the end of the list, not in the middle; better, list them from the start.
89
+ WARN session/unlock-coverage · unsure
90
+ unlock_toolset looks like it unlocks tools, and the run got through 1 of its 2 toolset values. […]
91
+ not unlocked: "reports"
92
+ → Next: Unlock every value: session --init puts each one in the starter.
187
93
  ```
188
94
 
189
- Or just the server, in any client: `claude mcp add toolmenu -- npx -y toolmenu mcp`, or
190
- `{"command": "npx", "args": ["-y", "toolmenu", "mcp"]}` in an `mcpServers` config.
95
+ Every finding has the same parts: a **level** (`ERROR`, `WARN`, `INFO`; `--fail-on`
96
+ decides which fail the run), a **rule id** you can look up (`toolmenu explain
97
+ session/mid-insert`) or turn off in the config, **where** it happened (the step, the
98
+ tool, the parameter path), **what toolmenu saw**, and a **`→ Next:`** step.
99
+
100
+ How much to trust a finding depends on what it rests on, and the rule id's prefix says which:
191
101
 
192
- | Tool | Runs |
102
+ | Rule family | What it rests on |
193
103
  |---|---|
194
- | `snapshot_menu` | `snapshot`: findings and token cost; writes the menu file only with `save_as` |
195
- | `diff_menus` | `diff`: breaking changes, the token change and the version bump between two menu files |
196
- | `check_session` | `session --auto`: calls the read-only tools and reports each menu change |
197
- | `explain_rule` | `explain`: what a rule checks and how to change its level |
198
-
199
- Each tool runs the CLI in a child process and returns its text report, so options,
200
- `toolmenu.config.json` and stored OAuth logins work as they do on the command line.
201
- Relative paths resolve from the folder `toolmenu mcp` runs in, or a call's `cwd`.
202
- `snapshot_menu` and `check_session` start the command you give them, as the CLI does;
203
- clients ask before each tool call unless you allow it. toolmenu's own menu passes its
204
- checks with no findings, on both protocols and in a session (the test suite checks).
205
-
206
- ## As a library
207
-
208
- Everything the CLI does is exported from the `toolmenu` package, with types, for test
209
- suites and scripts of your own:
210
-
211
- ```js
212
- import { snapshot, diffMenus, loadMenu, session, loadScenario } from 'toolmenu';
213
-
214
- // The menu and its findings, as `toolmenu snapshot` gives them.
215
- const { menu, findings } = await snapshot({ kind: 'stdio', command: 'node', args: ['dist/server.js'] });
216
- // or { kind: 'http', url: 'https://example.com/mcp', headers: { Authorization: `Bearer ${token}` } }
217
-
218
- // What changed since the committed baseline, and the version bump it needs.
219
- const diff = diffMenus(await loadMenu('menu.json'), menu);
220
- console.log(diff.suggestedBump, diff.tokens);
221
- if (diff.findings.some((f) => f.severity === 'error')) process.exitCode = 1;
222
- ```
104
+ | `spec/`, `apps/`, `lists/schema` | The protocol: the official JSON schema for the server's version (pinned in `schemas/`), a MUST/SHOULD in the spec, or the MCP Apps extension |
105
+ | `diff/` | Two menus compared. A **breaking** change is one where a call that passed validation before now fails it |
106
+ | `session/`, `menu/`, `lists/variance`, `auth/`, `catalog/` | Behaviour toolmenu observed: the menu changing after a step, between two calls, between two processes or connections, or served to a bad credential. The finding names the step or the call |
107
+ | `naming/`, `ids/`, `description/`, `write/`, `search/` | How agents read the menu: names, words, descriptions. Tuned on real servers, but heuristics |
223
108
 
224
- `snapshot`, `diffMenus`, `session` and `history` take the same options as their
225
- commands (`SnapshotOptions`, `DiffOptions`, `SessionOptions`, `HistoryOptions`) and
226
- return the findings the reports print: each `Finding` has its `rule`, `severity`,
227
- `message`, `tool` and the `fix` shown as `→ Next:`. Lower down, `compareMenus(before,
228
- after)` returns each tool's change (`ToolChange[]`) and `runRules` runs the menu rules
229
- on a menu you built. The `menu.json` format is the `Menu` type (`toolmenu: 1`);
230
- a later format will bump that number. Library functions don't print or exit: that's
231
- the CLI's job.
109
+ A finding marked **`· unsure`**, in any family, rests on names and words rather than on
110
+ something toolmenu saw happen (above: "looks like it unlocks tools"). It says what was
111
+ seen, not what it means. No LLM is involved anywhere: the same server gives the same
112
+ findings on every run.
232
113
 
233
114
  ## In CI: one PR comment
234
115
 
116
+ `npx toolmenu init -- <your server command>` snapshots the server into `menu.json` (the
117
+ baseline) and writes this workflow for your project (Node, Python, Go or Rust). Commit
118
+ both. Credentials you pass go into the workflow as `${{ secrets.… }}`, never as text.
119
+
235
120
  ```yaml
236
121
  # .github/workflows/toolmenu.yml
237
122
  on: pull_request
@@ -246,525 +131,127 @@ jobs:
246
131
  - uses: actions/setup-node@v7
247
132
  with: { node-version: 22 }
248
133
  - run: npm ci && npm run build
249
- - uses: niksa90/toolmenu@v0.25.2
134
+ - uses: niksa90/toolmenu@v0.27.1
250
135
  with:
251
136
  command: node dist/server.js
252
137
  baseline: menu.json # your committed snapshot
253
138
  scenario: scenario.yml # optional: run a session too
254
139
  ```
255
140
 
256
- It snapshots the PR's build, diffs it against the baseline as it is on the PR's base
257
- branch (or at the previous tag, on a tag push), runs the scenario, and
258
- posts **one comment that updates on every push**: breaking changes, the token change
259
- ("this PR adds ~1,240 tokens to every conversation"), the suggested version bump,
260
- and anything `session` caught. The same report goes to the job summary.
261
-
262
- | Input | Default | |
263
- |---|---|---|
264
- | `command` or `url` | | How to start the server over stdio, or its Streamable HTTP endpoint |
265
- | `headers` | | HTTP headers for a `url` server, one `Name: value` per line. Use a secret for keys: `x-api-key: ${{ secrets.MCP_API_KEY }}` |
266
- | `env` | | Environment variables for a `command` server, one `KEY=value` per line |
267
- | `baseline` | `menu.json` | The committed snapshot to diff against |
268
- | `scenario` | | A scenario to run with `session`, or `auto` to build the steps from the menu |
269
- | `catalog` | `false` | `true` also diffs the operations behind a search tool (`snapshot --catalog`) |
270
- | `baseline-from` | `snapshot` | `session` diffs every tool the scenario saw (the union menu), for servers whose tools appear after an unlock |
271
- | `release` | `auto` | Release versions for the bump check. `auto` reads `package.json`, `pyproject.toml` or `Cargo.toml` on the base branch and the head, or compares the tag with the previous one on a tag push. Set `"1.4.0..1.5.0"` yourself, or `off` |
272
- | `fail-on` | `error` | Fail the job on findings at or above this level |
273
- | `comment` | `true` | Post and update the PR comment |
274
- | `sarif` | | A folder to also write the findings to as SARIF, for code scanning (below) |
275
- | `version` | the Action's own | toolmenu version from npm: by default the one matching the Action tag you pinned (`latest` for the newest). The Action installs toolmenu from npm, so it needs a published version |
276
-
277
- ### In the Security tab: code scanning
278
-
279
- With `sarif`, findings also go to GitHub code scanning: alerts in the Security tab,
280
- each on the line of its tool in your `menu.json`, and a code scanning check on the pull
281
- request. GitHub shows an alert inline in the pull request only when that line changed in
282
- it, so most stay in the Security tab and the check. Run the workflow on pushes to your
283
- default branch too, so GitHub has a base to compare a pull request's alerts with. Code scanning is free for public repositories; private ones need GitHub Advanced
284
- Security.
285
-
286
- ```yaml
287
- on:
288
- pull_request:
289
- push:
290
- branches: [main] # the base GitHub compares a pull request's alerts with
291
- permissions:
292
- contents: read
293
- pull-requests: write
294
- security-events: write # to upload SARIF
295
- steps:
296
- # … checkout, build, as above
297
- - uses: niksa90/toolmenu@v0.25.2
298
- with:
299
- command: node dist/server.js
300
- sarif: toolmenu-sarif
301
- - uses: github/codeql-action/upload-sarif@v4
302
- if: always() # findings fail the toolmenu step; upload them anyway
303
- with:
304
- sarif_file: toolmenu-sarif
305
- ```
306
-
307
- Outside the Action: `toolmenu snapshot --sarif toolmenu.sarif -- node dist/server.js`
308
- writes the file beside the usual report (`diff` and `session` take `--sarif` too), or
309
- `--format sarif` prints it instead.
310
-
311
- ### Other CI systems: JUnit test reports
312
-
313
- GitLab, Jenkins, Azure DevOps, CircleCI and Buildkite show JUnit XML as test results.
314
- `--junit <path>` writes the findings of `snapshot`, `diff` or `session` as JUnit beside
315
- the usual report (`--format junit` prints it instead). New in 0.25.2. Each finding is a test case named
316
- after its tool, in a class named after its rule (`toolmenu.diff.diff/removed`), so the
317
- CI system tracks it across runs. Findings at or above `--fail-on` are failures, the rest
318
- pass with their text, so the test report agrees with the exit code. A clean run is one
319
- passing case.
320
-
321
- ```yaml
322
- # .gitlab-ci.yml
323
- toolmenu:
324
- image: node:22
325
- script:
326
- - npm ci && npm run build
327
- # Exit 1 means findings: keep going so both reports are written, then fail.
328
- - npx toolmenu snapshot --out new-menu.json --junit snapshot.xml -- node dist/server.js || code=$?
329
- - npx toolmenu diff menu.json new-menu.json --junit diff.xml || code=$?
330
- - exit ${code:-0}
331
- artifacts:
332
- when: always
333
- reports:
334
- junit: [snapshot.xml, diff.xml]
335
- ```
336
-
337
- ### A badge for your README
338
-
339
- ```sh
340
- npx toolmenu snapshot --badge toolmenu-badge.json -- node dist/server.js
341
- ```
342
-
343
- writes a [shields.io endpoint](https://shields.io/badges/endpoint-badge) file (tools,
344
- tokens, and errors or warnings, in red, yellow or green) and prints the README line
345
- for it. Commit it with `menu.json`: it changes when the baseline does, so nothing
346
- has to commit from CI. The badge at the top of this README is toolmenu's own MCP
347
- server (`toolmenu mcp`), checked by itself.
348
-
349
- Server behind auth, or PRs from forks and Dependabot? See [docs/github-action.md](https://github.com/niksa90/toolmenu/blob/main/docs/github-action.md).
350
- Server that unlocks tools on request? Snapshot the full menu for the baseline and run the session on the default one: [servers that unlock tools](https://github.com/niksa90/toolmenu/blob/main/docs/github-action.md#servers-that-unlock-tools).
351
- PRs without secrets are skipped, and a skipped check still passes; the `skipped` output tells you.
141
+ Every pull request then gets one comment that updates on each push. It leads with a
142
+ verdict (can this menu replace the baseline as it is?), then the breaking changes, the
143
+ token change, the version bump the change needs, and anything `session` caught:
352
144
 
353
- ## `session`: the menu changing while the agent works
354
-
355
- `session` runs a scripted session (the scenario *is* the agent), lists the menu
356
- after every step, and reports each change with the step that caused it.
357
-
358
- ```yaml
359
- # scenario.yml
360
- allow_writes: false # refuse to call tools not marked readOnlyHint
361
- steps:
362
- - list
363
- - call: search_forms
364
- args: { query: onboarding }
365
- - call: unlock_toolset
366
- args: { toolset: audits }
367
- - wait_for: tools_list_changed # optional; timeout_ms defaults to 5000
368
- ```
145
+ <p align="center">
146
+ <img src="https://raw.githubusercontent.com/niksa90/toolmenu/main/docs/review-comment.png" width="760" alt="toolmenu's review of two real releases of the official filesystem server, as markdown: don't accept as is, 1 breaking change (move_file is now destructive), suggested bump major, +24 tokens">
147
+ </p>
369
148
 
370
- `--init` writes a starter scenario from the live menu: it calls every read-only tool
371
- that needs no arguments, then repeats one. If a tool looks like it unlocks more tools
372
- (a `domains` or `toolset` parameter with an enum, "unlock" or "capabilities" in its
373
- name, a description that says it enables or loads tools), the starter unlocks
374
- **every** value, listing after each, then repeats the first, so the session sees every
375
- tool behind the unlock. The values come from the parameter's enum or, without one,
376
- from the server's own read-only listing (`list_toolsets`). Lookups that only describe
377
- toolsets (`list_toolsets`, `get_toolset_tools`) aren't unlocks, and one not marked
378
- read-only is suggested, never called. Anything it can't fill in is left as a
379
- commented-out step. An unlock nothing in the menu names, like joining a room that
380
- brings more tools, needs a step you add yourself. A scenario is only as good as its
381
- steps, so the starter is a floor, not a ceiling.
382
-
383
- On github-mcp-server 1.0.5 with `--dynamic-toolsets` (3 tools at the start, 19
384
- toolsets), unlocking every toolset reaches 107 tools: all 81 that `--toolsets all`
385
- serves, and 23 issue and pull-request tools that only the dynamic mode has.
386
-
387
- | Rule | Default | Catches |
388
- |---|---|---|
389
- | `session/mid-insert`, `session/reorder`, `session/remove`, `session/edit` | error | The menu changing mid-session anywhere but the end, with where the change starts and which value changed. Any change to the tool list invalidates the cached prompt: the whole tool list (its size, estimated) and the conversation after it are processed again |
390
- | `session/append` | warn | Tools added at the end of the list. Still a cache miss for the conversation when your client sends tools at the start of the prompt (Claude's Messages API does); cache-safe only when the client adds new tools after the cached content, as tool search (deferred loading) does |
391
- | `session/connection-local` | error on 2026-07-28 (HTTP) | A change only this connection sees. 2026-07-28: the tool set MUST NOT vary "per-connection or as a side effect of other requests on the connection" |
392
- | `session/side-effect` | warn on 2026-07-28 (stdio) | The same, where stdio can't tell per-connection from global |
393
- | `menu/process-variance`, `menu/connection-variance` | error | Checked before the first step, as in `snapshot` (below) |
394
- | `session/unannounced` | warn | The menu changed without `notifications/tools/list_changed`, although the server declared `listChanged` |
395
- | `session/untested` | warn, error if no call got through | Calls that failed before reaching the tool: authentication, something missing on this machine (no Chrome), the network. One finding for the run, not one per step, so an expired CI secret doesn't pass a run that tested nothing |
396
- | `session/refused`, `session/step-failed` | error | A write the scenario didn't allow, or a call that failed for another reason |
397
- | `session/tool-error` | warn | A tool that answered with an error for another reason (`isError`): the run tested less than it looks. The same error at several steps is one finding that names them |
398
- | `session/known-variance` | info | A mid-session change only in values that already vary from one tools/list to the next: same cause as the finding it names, not counted again |
399
- | `session/assumed-read-only` | info | Tools called on your word (`--assume-read-only`, `assume_read_only`) although the server doesn't mark them readOnlyHint |
400
- | `session/unlock-coverage` | warn | An unlock whose enum lists its values, called with only some of them: the tools behind the rest were never seen, so a baseline from the session misses them. With `--union-out` or `--auto`, an unlock never called at all too, with why `--auto` skipped it |
401
- | `session/nothing-called` | warn | `--auto` called no tools (all open-world, missing values, or not read-only): the run only listed the menu |
402
- | `session/session-lost` | error | The server ended the session after toolmenu opened a second connection with the same credentials (servers with one session per client; 2025-era servers only). The run stops there; rerun with `--processes 1` |
403
- | `session/handle-error` | warn | 2026-07-28: `--auto` called a read-only tool with a handle that can't exist (`toolmenu-unknown-session`), and it answered with a protocol error instead of a tool error the model can read |
404
- | `session/error-not-flagged` | warn | A tool answered with an HTTP error from the API behind it (`{"status":401,"object":"error",…}`) as a plain result, not marked `isError`, so a client that reads `isError` takes the call for a success (Notion's server). An auth one also counts toward `session/untested`. Tools that take a URL are left out: their answer is the page's (heuristic) |
405
- | `session/server-unresponsive` | error | The server stopped answering after a step: even `tools/list` timed out. The run stops there with this one finding, instead of every later step waiting out its timeout (a server that polls a device-code sign-in inside a tool call) |
406
- | `session/handle-roundtrip` | warn | With `--create-handles`: the tool that creates a session, basket or cart was called, and the tools that take it back refused the handle it had just returned (info when its result had no handle to find) |
407
- | `session/rate-limited` | error | The server kept refusing requests for being too many (a 429) after toolmenu waited about a minute, or a tool not marked `readOnlyHint` said it was rate-limited (it isn't called again). The run stops there, with the steps it didn't check; give the run its own server instance or raise the limit for it |
408
- | `session/scope-unchecked` | info | A second server process or connection couldn't start mid-session (a server that holds a file or a port), so whether a change was global wasn't checked. Said once for the run |
409
-
410
- **No scenario? `session --auto`** builds the steps from the menu: every read-only tool
411
- whose required arguments the schema itself gives (a `default`, `examples`, an `enum`,
412
- a type or format), cheapest first, then every value of each read-only unlock (outside
413
- `--max-calls`, and the first value twice: a second identical unlock should change
414
- nothing), then the first call again. It never guesses an ID: a tool that needs
415
- `owner`, `repo` or an issue key is skipped, and the report names the flag that fills
416
- it: `--value repo_path=/src/app` fills that parameter on every tool that requires it,
417
- `--value get_issue.issue_key=ABC-1` on one tool, `--values-file values.yml` takes a map
418
- of either. Servers that don't mark their tools `readOnlyHint` (DeepWiki, every
419
- 2024-11-05 server) get nothing called unless you vouch for them by name:
420
- `--assume-read-only read_wiki_structure,ask_wiki_question`. Only exact names are
421
- accepted: a pattern would also cover tools a later release adds. A tool marked or
422
- named as a write is never called, and the report says which tools ran on your word.
423
- With `--create-handles` on 2026-07-28, `--auto` also calls the tool that creates a
424
- session, basket or cart (`start_session`; a write, so only with the flag, and never
425
- one marked `destructiveHint`; not locks, leases or checkouts) and gives the tools that
426
- take it back the handle it returned, beside the made-up one.
427
- `--save-scenario auto.yml` writes the steps, with the skipped tools commented out, for
428
- you to fill in. Tools marked `openWorldHint: true` (web
429
- search, fetch, scraping, which can cost API credits) are called only with
430
- `--open-world`, except unlocks: changing the menu is what a session watches, and an
431
- unlock spends no search credits, so it runs even on a server that marks every tool
432
- open-world. A run that ends up calling nothing is `session/nothing-called` (warn),
433
- with why each tool was left out, not a clean pass. On the 22 servers in `bench/`,
434
- `--auto` reaches 60 of 262 read-only tools, 84 with `--open-world`.
435
-
436
- **`session` calls tools for real.** Without `allow_writes: true` it refuses any tool
437
- not marked `readOnlyHint: true`. `--plan` prints the steps without connecting.
438
-
439
- ## `diff`: compare releases
149
+ That's `toolmenu review --format markdown` on two real releases of the official
150
+ filesystem server. The same findings go to GitHub code scanning (SARIF), to GitLab,
151
+ Jenkins and others as JUnit, or into one offline HTML page:
440
152
 
441
- ```sh
442
- npx toolmenu diff menu.json new-menu.json
443
- ```
153
+ <p align="center">
154
+ <img src="https://raw.githubusercontent.com/niksa90/toolmenu/main/docs/report.png" width="760" alt="toolmenu's HTML report of a session: 3 errors, 2 warnings, the menu after each step with inserted tools marked, and each finding with its next step">
155
+ </p>
444
156
 
445
- ```
446
- $ npx toolmenu diff --release 2026.1.14..2026.8.31 fs-2026.1.14.json fs-2026.8.31.json
447
- toolmenu diff secure-filesystem-server 2026.1.14 → 2026.8.31
448
- changes 1 breaking (1 tool) · 0 minor · 3 notice (13 tools)
449
- version suggested bump: major · not checked (calendar version)
450
- tokens ~1,640 → ~1,664 (+24, estimate): this release adds ~24 tokens to every conversation that loads the menu
451
- tools 14 → 14
452
-
453
- ERROR diff/safety-hint
454
- `move_file` was marked additive-only and is now destructive (destructiveHint false → true; unset means destructive). Clients that auto-approve additive tools may now ask first.
455
- → Next: If the tool still never deletes or overwrites, set destructiveHint: false; if it does now, say so in the release notes.
456
- INFO diff/annotations
457
- Annotations changed the same way (openWorldHint (unset) → false) in 13 of 14 tools: read_file, read_text_file, read_media_file, read_multiple_files, write_file, edit_file, … (+7).
458
- …
459
- ```
157
+ Inputs, servers behind auth, forks and `gate: review` are in
158
+ [docs/github-action.md](https://github.com/niksa90/toolmenu/blob/main/docs/github-action.md); SARIF, JUnit, HTML and the badge in
159
+ [docs/ci.md](https://github.com/niksa90/toolmenu/blob/main/docs/ci.md).
460
160
 
461
- That's two real releases of the official filesystem server. Its calendar versions
462
- promise nothing about compatibility, so the bump isn't judged, but the breaking change
463
- is still an error. Changes are classified
464
- the way OpenAPI breaking-change checkers do it:
161
+ ## Commands
465
162
 
466
- | Class | Changes | Rules |
163
+ | Command | What it does | |
467
164
  |---|---|---|
468
- | **breaking** (error) | tool removed or renamed, parameter removed, new required parameter, narrower or different type, enum narrowed, unlisted properties now rejected (`additionalProperties: false` added), a tool becoming less safe (no longer read-only, or additive → destructive); with `--lists` on both sides, a prompt removed or newly requiring an argument, a resource template removed | `diff/tool-removed`, `diff/tool-renamed`, `diff/param-removed`, `diff/param-required`, `diff/param-type`, `diff/enum-narrowed`, `diff/properties-closed`, `diff/safety-hint`, `diff/prompt-removed`, `diff/prompt-argument-required`, `diff/resource-template-removed` |
469
- | **minor** (info) | tool added, new optional parameter, parameter now optional, type or enum widened, unlisted properties now accepted (`additionalProperties: false` removed), a prompt or resource template added | `diff/tool-added`, `diff/param-added`, `diff/param-relaxed`, `diff/type-widened`, `diff/enum-widened`, `diff/properties-opened`, `diff/prompt-added`, `diff/resource-template-added` |
470
- | **notice** (info) | optional parameter dropped (extra properties still allowed), description changed (with the text diff: it changes what the agent does), output schema, annotations, other fields, order, a schema restructured but accepting the same input, only the declared `$schema` dialect changed, a prompt's description or optional arguments, resources added or gone (data, so no bump) | `diff/param-dropped`, `diff/description`, `diff/schema-other`, `diff/annotations`, `diff/other`, `diff/order`, `diff/schema-equivalent`, `diff/schema-dialect`, `diff/prompt-changed`, `diff/resources` |
471
-
472
- The same rules apply inside parameters: fields of an object parameter, array items,
473
- and the options of an `anyOf`/`oneOf` union (zod's unions, discriminated unions and
474
- `.nullable()`) are compared at every depth, with their path (`gen.body.text`,
475
- `search.filters[].field`, `gen.block(kind="image").url`). Union options are paired by
476
- the value of a discriminator they all fix; without one (a plain `z.union` of objects),
477
- identical options first, so a reorder is no change, then by type and shared property
478
- names. An option gone is breaking, a new one widens. The same change reached through
479
- one shared definition at several places is one finding that lists them (`places` in
480
- `--json`). Local `$ref`s (`#/$defs/…`, `#/definitions/…`) are expanded first,
481
- recursive ones one level deep, so a schema is compared by what it accepts, not how
482
- it's spelled: moving a repeated block into `$defs` is one `diff/schema-equivalent`
483
- notice with its token change ("accepts the same input: ~899 tokens fewer"), and a
484
- breaking change inside `$defs` is breaking. A schema change no rule classifies is
485
- never silent: it's a `diff/schema-other` "review it".
486
-
487
- On mongodb-mcp-server 2.1.2 → 3.0.0 this finds `aggregate.pipeline[]…$vectorSearch.numCandidates`
488
- and `.limit` narrowed from `number` to `integer` inside a union, where the whole
489
- `pipeline` used to be one "review it".
490
-
491
- Different spellings of the same schema are one schema (`additionalProperties` absent,
492
- `true` or `{}`; zod 4's safe-integer bounds), and a change every tool shares is said
493
- once for the menu: a server moving from zod 3 to zod 4 reads as its real changes (a
494
- field that became required, the patterns zod 4 adds to `email` and `uuid`) plus one
495
- `diff/properties-opened` line ("27 tools now accept properties they don't list … most
496
- likely a schema generator upgrade"), not a "review it" per object.
497
-
498
- The same change in many tools is reported once, with the tools it touches (`pageId`
499
- is new and required in 25 of 29 tools, on chrome-devtools-mcp 1.8.0); counts are per
500
- change. When a tool loses a parameter and gains a required one with a close name,
501
- `diff/param-renamed` (warn, unsure) points out the likely rename; both changes stay
502
- breaking. Every breaking finding's next step names the version to release when the
503
- bump is too small.
504
-
505
- It also reports the token change per tool and suggests a semver bump. To check the
506
- bump, pass the **release** versions: `--release 1.4.0..1.5.0` (npm, a git tag).
507
- `diff/version-bump` then warns when the release's bump is smaller. Under 1.0.0 it
508
- reads versions as npm's caret does (`^0.2.3` accepts any 0.2.x): breaking changes need
509
- a minor bump, new features a patch, and under 0.1.0 anything goes.
510
- `diff/version-backwards` warns when the version goes down. Calendar versions and prereleases aren't judged. The version in the snapshot is what the
511
- server reports (`serverInfo.version`), which is often not the release: the filesystem
512
- server has said `0.2.0` for 19 releases (FINDINGS F5). It's shown, and only checked
513
- with `--server-version-is-release`. `diff` also enforces an optional `tokenBudget`
514
- (`diff/token-budget`). Order changes are only a notice here: between releases, prompt caches rebuild. Order
515
- matters *within* a session.
516
-
517
- ### Operations behind a search tool (`--catalog`)
518
-
519
- Big servers increasingly keep most operations out of the menu: a small fixed menu, a
520
- search tool, and an execute tool (Sentry's `search_sentry_tools` + `execute_sentry_tool`,
521
- Atlassian's `discover` + `executeRead`). A breaking change to one of those operations
522
- never shows in the menu. `snapshot --catalog` asks the search tool a fixed set of
523
- queries and keeps the operations it returns in `menu.json`; `diff` then compares
524
- operations found in both snapshots with the same rules as tools.
525
-
526
- ```sh
527
- npx toolmenu snapshot --catalog -- node dist/server.js
528
- ```
529
-
530
- A search returns only its top matches, so toolmenu crawls the catalog by its own
531
- words: it starts from the search tool's own example phrases and the nouns in your tool
532
- names, then searches for every operation it finds or sees mentioned, until 20 queries
533
- in a row turn up nothing new (at most 200). On Atlassian's hosted server that finds all
534
- 154 operations behind its 21 tools; on Sentry's, 65 behind 9, the same on every run.
535
- Queries are paced and a rate limit is waited out. An operation not found this time is a
536
- notice (`diff/catalog-missing`: "not returned by the same queries"), never "removed";
537
- one found only now is `diff/catalog-added`, and catalogs read with different queries,
538
- or partly (a query failed), are `diff/catalog-queries`. To steer it, set
539
- `"catalog": { "queries": [...], "maxQueries": 200, "crawl": true }` in
540
- `toolmenu.config.json` (FINDINGS F15).
541
-
542
- A crawl through search is a lower bound, not a proof: an operation that nothing in the
543
- catalog names, and whose words no query uses, stays unfound. On `@sentry/mcp-server`
544
- 0.42 (stdio) it finds 63 of the 64 operations in the package's catalog; `whoami` comes
545
- back only when asked for by name. Add queries like that to `catalog.queries`.
546
-
547
- **Command routers.** Some servers put their operations behind router tools instead:
548
- one tool per area that takes a `command` and its `parameters`, and lists its commands
549
- when called with `learn: true` (Azure's MCP server serves 71 such tools in front of
550
- ~410 operations). `--catalog` detects them (a `command` parameter, an arguments
551
- object, a boolean listing flag, and a description that says "router" or "sub
552
- commands") and calls each once in its listing mode, never with a command, so nothing
553
- runs. Routers whose command is required aren't called. Detection missed one? Name
554
- them: `"catalog": { "routers": ["keyvault", "storage"] }`. A router lists everything
555
- it has, so unlike a search the catalog is complete except for routers that failed
556
- (`catalog/failed` says which, and why).
557
-
558
- ## `history`: a package's releases, researched
559
-
560
- ```sh
561
- npx toolmenu history @modelcontextprotocol/server-filesystem --versions 19 --arg /tmp --csv fs.csv
562
- ```
563
-
564
- It installs each published version into a temp directory (install scripts off),
565
- snapshots it, and diffs it against the previous version that worked. For every
566
- version it records the protocol, the menu, the findings and the **dependency
567
- versions that actually got installed**: an old version installed today resolves its
568
- dependencies today, and can serve a different menu than it shipped with.
569
-
570
- It's best effort: packages that need env vars, arguments or auth are recorded as
571
- failed, with a reason, never skipped. **It installs and runs third-party code. Run it
572
- in a container.**
573
-
574
- ## Where the tokens go
575
-
576
- `snapshot` also shows what the menu's tokens are spent on: the biggest tools (split
577
- into description and schema), large enums, and waste you can remove:
578
-
579
- - **Repeated blocks**: the same piece of schema at several places, inside one tool (a
580
- `$defs` entry could hold it once: "one ~230-token block ×5 in
581
- content_generateMessageHtml") or across tools (paid once per tool: mongodb-mcp-server
582
- 3.0.4 repeats a ~2,200-token `pipeline` schema in 3 tools). Only the largest repeated
583
- block is named, not the pieces inside it.
584
- - **Unused `$defs`**: definitions nothing in the tool refers to. Notion's server 2.5.2
585
- ships the same 9 in all 24 tools, 72% of its menu.
586
-
587
- Every tool's estimate is in the menu file, for the ones the summary leaves out:
588
- `jq -r '.tools[] | "\(.tokens)\t\(.name)"' menu.json | sort -rn`.
589
-
590
- ## `snapshot` rules
591
-
592
- Every rule comes from a failure I actually hit building 115 tools for one MCP
593
- server, or from the spec itself. If a rule can't point to one, it doesn't ship.
594
-
595
- | Rule | Default | Catches |
165
+ | `snapshot` | Connects, lists the menu, checks it (the spec, a second process, your client's limits, token cost) and writes `menu.json` | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/snapshot.md) |
166
+ | `session` | Plays a scenario, or builds one with `--auto`: calls tools, checks every result, and lists the menu after each step. Read-only tools unless you allow writes; `--call-everything` calls every tool, for a server in a sandbox | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/session.md) |
167
+ | `diff` | Compares two menu files: breaking, minor, notice; token change; the semver bump | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/diff.md) |
168
+ | `review` | Should the new menu replace the baseline? One verdict and an exit code | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/diff.md#accepting-a-new-baseline-review) |
169
+ | `auth check` | Walks an HTTP server's OAuth discovery as a spec-following client does, without logging in | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/usage.md#http-servers-and-oauth) |
170
+ | `auth login` | Signs in to a server behind OAuth once, in a browser; `snapshot` and `session` then use that login | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/usage.md#http-servers-and-oauth) |
171
+ | `scan` | `snapshot` on every server in a client config (Claude Desktop, `.mcp.json`, Cursor, VS Code), side by side | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/snapshot.md#every-server-you-use-scan) |
172
+ | `history` | Installs a package's published versions one by one and diffs them. Runs third-party code: use a container | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/diff.md#history-a-packages-releases-researched) |
173
+ | `init` | Writes the baseline and the CI workflow | [above](#in-ci-one-pr-comment) |
174
+ | `mcp` | toolmenu as an MCP server and a Claude Code plugin, so the agent building your server can check it while it works | [docs](https://github.com/niksa90/toolmenu/blob/main/docs/usage.md#for-the-agent-building-your-server-mcp) |
175
+ | `explain <rule>` | What a rule checks, its default level and how to change it | [RULES](https://github.com/niksa90/toolmenu/blob/main/docs/RULES.md) |
176
+
177
+ `--client <name>` checks the menu for one client, from that client's own docs: Claude
178
+ Code, the Claude API, OpenAI, Codex, VS Code (Copilot Chat), Copilot cloud agent and code
179
+ review, Gemini CLI, the Gemini API or Antigravity ([docs](https://github.com/niksa90/toolmenu/blob/main/docs/clients.md)). A client
180
+ that doesn't document a limit doesn't get one: Cursor's and Windsurf's aren't checked.
181
+
182
+ `session` calls tools for real. Without `allow_writes: true` it refuses any tool not
183
+ marked `readOnlyHint: true`, and `--call-everything` is refused over HTTP, because a
184
+ hosted server's tools act on someone's real account.
185
+
186
+ Everything is also exported as a library, with types
187
+ ([docs](https://github.com/niksa90/toolmenu/blob/main/docs/usage.md#as-a-library)). All protocol versions from 2024-11-05 to
188
+ 2026-07-28 are supported, through the official TypeScript SDK; the table in
189
+ [docs/PROTOCOLS.md](https://github.com/niksa90/toolmenu/blob/main/docs/PROTOCOLS.md) is written by a test that runs each one.
190
+
191
+ ## Tested on real servers
192
+
193
+ Rules are checked against public MCP servers before they ship, and a rule that changes
194
+ a finding on one has to say why in the [changelog](CHANGELOG.md). The run on 2026-11-09
195
+ is the baseline for a weekly one ([`bench/scripts/weekly.sh`](bench/scripts/weekly.sh)),
196
+ which compares each week with the last:
197
+
198
+ | Panel | Servers | What ran |
596
199
  |---|---|---|
597
- | `menu/nondeterministic` | error | Two identical `tools/list` calls returning different menus (order, descriptions, schemas, annotations), with the value that differs |
598
- | `menu/process-variance` | error | A second server process, started the same way, serves a different menu, so every restart misses the cache. Seen on mcp-atlassian: a default built from a Python set, in a new order every start (FINDINGS F12). toolmenu runs the main process with `PYTHONHASHSEED=0`, so a Python server's saved menu is reproducible, and the second with another seed. `--processes 1` turns it off |
599
- | `menu/connection-variance` | error (warn on 2025 protocols) | The same over HTTP: a second connection with the same credentials gets a different menu. Retried once before reporting |
600
- | `spec/schema` | error | A `tools/list` result that fails the official schema for its protocol version |
601
- | `spec/tools-capability` | error | The server doesn't declare the `tools` capability, so clients (the official SDK included) never ask for its tools |
602
- | `menu/duplicate-name` | error | Two tools with the same name: only one can be called, and some clients reject the list |
603
- | `naming/route` | error | A `routes.yml` expectation broke: a keyword now matches the wrong tool at least as well as the right one |
604
- | `description/buried` | warn | Instructions to the agent ("use X instead", "don't retry", "never guess one") past the point where your client cuts descriptions: in Claude Code 16,384 characters for a tool loaded through tool search (its default) and 4,096 for one that loads up front, or your `descriptionLimit`. From a real failure: a client cut at 280 characters, and the line that decided routing was at 1,222 |
605
- | `description/cut` | info | Descriptions longer than the client sends, with nothing that reads as an instruction past the cut, as one summary that shows where each one is cut. The model gets a prefix that can read as complete |
606
- | `description/late-instruction` | info | With the default cut: instructions after character 280, which a client that cuts descriptions short would hide, as one summary. Set `descriptionLimit` and `description/buried` checks your real cut instead. (Before 0.13 this was part of `description/cut`; that id's setting still applies.) |
607
- | `naming/vague-id` | warn | Parameters called just `id` that don't say *which* thing, as one summary with a suggested name |
608
- | `ids/authored` | info | An ID the agent must supply that no tool appears to return, so the agent may invent it, one finding per kind of ID (heuristic: IDs from a URL or another tool's text output look the same) |
609
- | `ids/handle-lifetime` | info | 2026-07-28: a tool that creates a handle another tool takes back (`start_session` → `session_id`) doesn't say how long it lasts (heuristic) |
610
- | `write/unannotated` | warn | Tools named like writes (`delete_`, `send_`…) with no annotations, as one summary grouped by the annotation each name suggests |
611
- | `naming/shared-word` | info | Nouns that name different things across tools (`list_team_audits` vs `get_audit_trail`), as one summary |
612
- | `write/read-only-mismatch` | warn | A tool marked `readOnlyHint: true` whose name says it writes (`create_my_account`): clients act on the hint, and `--auto` doesn't call it on the hint alone (heuristic) |
613
- | `write/no-dry-run` | info | Destructive tools with no dry-run parameter or preview tool, as one summary |
614
- | `spec/discover` | info | The server only speaks the 2025 protocol, so 2026-07-28 rules are skipped |
615
- | `spec/deprecated`, `spec/cache-hints` | info | 2026-07-28: deprecated features still advertised; `ttlMs: 0`, or `cacheScope: "public"` on a server that took credentials |
616
- | `lists/variance` | warn | With `--lists`: `prompts/list`, `resources/list` or `resources/templates/list` differs between two calls, or from a second process or connection (2026-07-28: lists no longer vary per connection) |
617
- | `lists/schema` | error | With `--lists`: a list result that doesn't match the official schema (2026-07-28 requires `ttlMs` and `cacheScope` on each), or a declared list that fails to answer (warn; info for Method not found, or a setup failure such as missing credentials) |
618
- | `naming/client-name` | warn | With `--client`: tool names the client's API refuses: `^[a-zA-Z0-9_-]{1,128}$` for the Claude API, `^[a-zA-Z0-9_-]+$` up to 128 for OpenAI's functions. MCP allows names such as `files.read` that neither takes as they are |
619
- | `menu/auth-variance` | info | With `--as` (or `identities` in the config): the tool set differs by identity, what each adds, lacks or defines differently against the first, and its tokens. Allowed by the spec; a baseline of one identity's menu doesn't cover the others. An identity that can't be listed is said here too |
620
- | `spec/cache-scope-public` | warn | 2026-07-28, with `--as`: the tool set differs by identity, and `tools/list` says `cacheScope: "public"`, which lets shared caches serve one caller's list to another |
621
- | `menu/capability-dependent` | info | Tools a server lists only to a client that supports elicitation, or only to one that doesn't. `snapshot` lists as a client with elicitation, so the menu file holds those tools (github-mcp-server 1.14 lists `delete_repository` only to a client that can ask the user to confirm; before 0.22 a baseline missed it), and once more without |
622
- | `menu/network-dependent` | warn | With `--offline-check` (stdio): the menu listed once more with the network cut off (the proxy variables pointed at a closed port, loopback left alone) differs: inite-visibility serves 3 tools offline and 10 online, so a baseline made with the network doesn't match a machine without it. A server that doesn't start offline is info. A server that ignores the proxy variables still reaches the network where the machine has it directly; then nothing is reported, which isn't a pass |
623
- | `auth/invalid-accepted` | warn | HTTP with credentials: the menu listed once more with an invalid credential in the same scheme. A server that serves a menu to it, not 401, gives a client whose token expired no reason to log in again; the agent just has fewer tools (Atlassian Rovo served 4 tools to any `Authorization` header). `--processes 1` turns it off |
624
-
625
- The full list, `diff` and `session` rules included, is [docs/RULES.md](docs/RULES.md).
626
-
627
- The naming and ID rules are heuristics, tuned on real servers (FINDINGS F7 and
628
- F10) and a private 115-tool one. They're the extras; the change checks are the point.
629
-
630
- **Pin routing with `routes.yml`.** Two tools fighting over one word needs a regression
631
- test, not a rename you'll forget:
632
-
633
- ```yaml
634
- audit:
635
- must_match: [list_team_audits]
636
- must_not_match: [get_audit_trail]
637
- audit trail:
638
- must_not_match: [get_store_scores] # on its own: must not match at all
639
- ```
640
-
641
- Matching is keyword matching, the way a keyword or BM25 tool search sees the menu: a
642
- tool matches if every word of the route is in its name or description (the exact
643
- phrase scores higher). It can't read "not", and neither can those searches: a
644
- disclaimer like "not the audit trail" puts "audit trail" into exactly the tool you
645
- meant to steer away from, and toolmenu says so.
646
-
647
- **Config** (optional `toolmenu.config.json`):
648
-
649
- ```json
650
- {
651
- "rules": { "naming/shared-word": "off", "ids/authored": "error" },
652
- "ignore": ["debug_*"],
653
- "routes": "routes.yml",
654
- "tokenBudget": 4000,
655
- "descriptionLimit": "claude-code",
656
- "fullDescriptions": ["search_capabilities", "lookup_*"],
657
- "offlineCheck": true
658
- }
659
- ```
660
-
661
- `offlineCheck` turns on `--offline-check` for every `snapshot` of a stdio server.
662
-
663
- `descriptionLimit` is where your client cuts tool descriptions: a number, or a client
664
- whose cut has been reported: `"claude-code"` (the default: Claude Code's own docs,
665
- [code.claude.com/docs/en/mcp](https://code.claude.com/docs/en/mcp), read 2026-11-02:
666
- 16,384 characters for a tool loaded through tool search, which is on by default, and
667
- 4,096 for a tool that loads up front, marked `_meta["anthropic/alwaysLoad"]: true`;
668
- the cut was 2,048 before v2.1.296), `"claude-code-upfront"` (4,096 for every tool:
669
- Claude Code with tool search off) or
670
- `"amazon-q"` (10,024, reported in
671
- [makenotion/notion-mcp-server#145](https://github.com/makenotion/notion-mcp-server/issues/145)).
672
- Clients that send descriptions in full don't need a setting beyond that. If your own
673
- client cuts shorter, as mine did at 280, set its number. `fullDescriptions` lists the
674
- tools it sends uncut, so the description rules skip them.
200
+ | Registry stdio servers (npm), pinned versions | 1,089 | `snapshot`, `session --auto`, and every tool called in a container with no network and no credentials: 15,550 of 16,518 tools |
201
+ | Docker's MCP catalog, at the digests it pins | 237 | The same: 4,117 of 4,237 tools called (41 servers didn't list tools) |
202
+ | Servers vendors publish to run locally (GitHub, AWS Labs, Microsoft, PayPal, Stripe, Notion, HashiCorp, Grafana…) | 98 | The same, with dummy credentials: 2,213 of 2,292 tools called (20 didn't list tools) |
203
+ | Hosted servers: 109 vendor endpoints and a seeded 2,000 from the registry | 2,109 | `snapshot`, a read-only `session --auto` of at most 10 calls, `auth check` on each that answers 401 |
675
204
 
676
- ### Checking for one client (`--client`)
205
+ Read those numbers for what they are. Most servers pass most checks: about 1 in 90 of
206
+ the registry's stdio servers fails in the official SDK client. A sealed container with
207
+ made-up input shows whether a tool crashes, hangs or breaks its own schema, not whether
208
+ it does its job: more than half of the calls there came back as a readable tool error,
209
+ the right answer to made-up input or a dummy key. Hosted servers got a few read-only calls each.
677
210
 
678
- What an agent sees depends on its client. `snapshot --client <name>` (or `"client"` in the
679
- config) checks the menu against what that client's own docs say, and every value
680
- toolmenu uses names its source and the date it was read (`src/clients.ts`). What a
681
- client doesn't document is left out, not guessed.
211
+ The reference server, `@modelcontextprotocol/server-everything`, gets no false finding,
212
+ and a clean fixture server must get none from any command; the test suite fails if one
213
+ appears. What the runs found, including what toolmenu itself got wrong and how it was
214
+ fixed, is in [docs/FINDINGS.md](https://github.com/niksa90/toolmenu/blob/main/docs/FINDINGS.md). The corpora, the scripts and the
215
+ datasets are in [bench/](bench/README.md).
682
216
 
683
- | `--client` | Up front | Search | Description cut | Names |
684
- |---|---|---|---|---|
685
- | `claude-code` | tool names; full definitions for tools marked `_meta["anthropic/alwaysLoad"]` | reads a description up to 4,096 characters | 16,384 through search, 4,096 up front | Claude API's |
686
- | `claude-api-search` | nothing from tools you defer (keep the 3–5 used most loaded) | name, description, argument names and descriptions; 5 results by default | none documented | `^[a-zA-Z0-9_-]{1,128}$` |
687
- | `openai` | each deferred function's name and description (its schema loads on demand); a deferred MCP server shows only its own name and description | how many results, and which fields, isn't documented, so `search/findable` doesn't run | none documented | `^[a-zA-Z0-9_-]+$`, up to 128 |
688
-
689
- The snapshot adds a line such as `as Claude Code: up front ~524 tokens · on demand ~344
690
- per tool a search loads (largest: update_contacts, ~2,093)`, and `--json` a `client`
691
- object. With a client that defers tools, `search/findable` runs at any menu size.
692
-
693
- ## What toolmenu can't catch
217
+ ## What it can't catch
694
218
 
695
219
  toolmenu sees what the server sends. Plenty of agent failures happen elsewhere:
696
220
 
697
- - **Routing to a feature with no tools.** If the agent reaches for something your
698
- menu doesn't have, no menu check can see the gap.
699
- - **What the client does with the menu.** Truncating descriptions, sorting or
700
- filtering tools, merging several servers. The description rules use Claude Code's
701
- reported cut unless you set yours (`descriptionLimit`); they can't see your client.
702
- - **Your code and your logs.** A capability check that's answered with one yes/no
703
- for two different behaviours, or a log line that reads the same whether something
704
- happened or not. Those live in the client or the server's code.
705
- - **Whether the agent actually picks the right tool.** That needs evals with a model.
706
- toolmenu stays deterministic on purpose.
707
-
708
- ## Options
709
-
710
- ```
711
- --out <path> where to write the menu (default: menu.json)
712
- --no-write don't write the menu file
713
- --routes <path> routes.yml
714
- --config <path> config file (default: toolmenu.config.json, if present)
715
- --format <fmt> text (default), json, github (annotations), markdown, sarif
716
- (code scanning) or junit (CI test reports)
717
- --junit <path> snapshot, diff, session: also write the findings as JUnit XML
718
- --fail-on <level> error (default), warn or info
719
- --header <k: v> HTTP header, repeatable
720
- --env <K=V> env var for a stdio server, repeatable (it only gets a
721
- minimal environment otherwise)
722
- --timeout <ms> per-request timeout (default: 30000)
723
- --protocol <p> auto (default), legacy (2025 handshake only) or modern
724
- (2026-07-28 only; an error if the server lacks it)
725
- --processes <n> server processes or connections to compare (default: 2);
726
- 1 opens no second one, session's scope check included
727
- --lists <kinds> snapshot: also read and check prompts, resources (comma-separated);
728
- saved in the menu file beside the tools
729
- --client <name> snapshot: check the menu for one client, from its own docs:
730
- claude-code, claude-api-search or openai
731
- --as <name>:<cred> snapshot: list the menu as this identity too, repeatable:
732
- "reader:Authorization: Bearer …" (HTTP) or "admin:ROLE=admin"
733
- (stdio). The first is the menu saved; the others are compared
734
- --offline-check snapshot (stdio): also list the menu with the network cut off
735
- --concurrency <n> scan: servers checked at a time (default: 4)
736
- --out-dir <dir> scan: write each server's menu file there, as <name>.json
737
- ```
738
-
739
- Exit codes: `0` clean · `1` findings at or above `--fail-on` · `2` couldn't connect
740
- (for `scan`: a server wasn't checked) or bad usage. `toolmenu <command> --help` lists that command's own options.
741
-
742
- ## Limitations
743
-
744
- - **Server order isn't final client order.** Clients merge tools from several servers
745
- and may sort or filter them. A stable server menu is necessary for cache hits, not
746
- sufficient. toolmenu measures what the server controls.
747
- - **Token counts are estimates** (`o200k_base`) of what the model reads: each tool's
748
- name, description and input schema. Output schemas, annotations, icons and `_meta`
749
- stay with the client and aren't counted. Vendors tokenize, cache and bill
750
- differently. `diff` recounts both menus with the toolmenu that runs it, so a
751
- baseline from an older version still compares fairly; a number quoted elsewhere (a
752
- commit message, a dashboard) should say which toolmenu counted it (0.7.1 changed the
753
- count for menus with icons by up to 5×, FINDINGS F7).
754
- - **Heuristic rules say so** in their messages.
755
- - Not a security scanner or a full conformance suite. Other tools do those well.
221
+ - **Whether the agent picks the right tool.** That needs evals with a model. toolmenu
222
+ stays deterministic on purpose; `routes.yml` pins keyword routing, which is as far as
223
+ it goes.
224
+ - **What the client does with the menu.** Clients merge several servers, sort, filter
225
+ and cut. A stable server menu is necessary for cache hits, not sufficient. The
226
+ `--client` profiles cover what each client documents, and nothing it doesn't.
227
+ - **A feature with no tools.** If the agent reaches for something your menu doesn't
228
+ have, no menu check sees the gap.
229
+ - **Whether a tool does its job.** A call that answers is checked against the schema it
230
+ declares, not against what it should have done.
231
+ - **Security.** It isn't a security scanner or a full conformance suite.
232
+
233
+ Token counts are estimates (`o200k_base`) of what the model reads: each tool's name,
234
+ description and input schema. Vendors tokenize, cache and bill differently.
235
+
236
+ ## Docs
237
+
238
+ - [snapshot](https://github.com/niksa90/toolmenu/blob/main/docs/snapshot.md): rules, where the tokens go, `routes.yml`, the config file, `scan`
239
+ - [session](https://github.com/niksa90/toolmenu/blob/main/docs/session.md): scenarios, `--init`, `--auto`, `--call-everything`, every session rule
240
+ - [diff, review, catalogs, history](https://github.com/niksa90/toolmenu/blob/main/docs/diff.md): how changes are classified, the verdicts, operations behind a search tool
241
+ - [clients](https://github.com/niksa90/toolmenu/blob/main/docs/clients.md): what each `--client` profile checks, and its sources
242
+ - [usage](https://github.com/niksa90/toolmenu/blob/main/docs/usage.md): HTTP and OAuth, proxies, the MCP server and plugin, the library, every option
243
+ - [the GitHub Action](https://github.com/niksa90/toolmenu/blob/main/docs/github-action.md) and [other CI output](https://github.com/niksa90/toolmenu/blob/main/docs/ci.md)
244
+ - [RULES](https://github.com/niksa90/toolmenu/blob/main/docs/RULES.md), [PROTOCOLS](https://github.com/niksa90/toolmenu/blob/main/docs/PROTOCOLS.md), [FINDINGS](https://github.com/niksa90/toolmenu/blob/main/docs/FINDINGS.md), [SPEC](https://github.com/niksa90/toolmenu/blob/main/docs/SPEC.md), [CHANGELOG](CHANGELOG.md)
756
245
 
757
246
  ## About
758
247
 
759
- Built by [Nikša Perović](https://niksa.me) while running a 115-tool MCP server and watching
760
- agents trip over its menu. Every rule started as one of those failures, and the
761
- research behind them is in [docs/FINDINGS.md](https://github.com/niksa90/toolmenu/blob/main/docs/FINDINGS.md). The story of building it, and where it was wrong, is [on niksa.me](https://niksa.me/toolmenu).
248
+ Built by [Nikša Perović](https://niksa.me) while running a 115-tool MCP server and
249
+ watching agents trip over its menu. It started as a check for menus that change at the
250
+ wrong time ([the story, and where it was wrong](https://niksa.me/toolmenu)) and grew
251
+ into a client-side test for the whole server as real servers kept failing in other
252
+ places.
762
253
 
763
254
  Found a false positive, or a failure toolmenu should catch?
764
255
  [Open an issue](https://github.com/niksa90/toolmenu/issues) with the menu (a
765
- `menu.json` is enough) and what you expected. Changes are listed in
766
- [CHANGELOG.md](https://github.com/niksa90/toolmenu/blob/main/CHANGELOG.md).
256
+ `menu.json` is enough) and what you expected.
767
257
 
768
- ## License
769
-
770
- MIT