# since-cutoff (pypi · since-cutoff) Which Python library APIs changed after your model's training cutoff (static diff, no model calls) - Trust score: 81/100 (high trust) - Registry status: active - Liveness: live - Owner verified: no - Last scored: 2026-10-02 ## Components - pypi · `since-cutoff`: 81/100 (this document), [markdown](https://verifymcp.io/servers/mohammadhijjawi97-since-cutoff/since-cutoff.md), [page](https://verifymcp.io/servers/mohammadhijjawi97-since-cutoff/since-cutoff) ## Channel facts - Registry: `pypi` - Package: `since-cutoff` - Version: `0.5.0` - Transport: `stdio` ## Trust breakdown How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog) Scored 2026-10-02. - **Supply Chain Security**: 100/100 - No malware found by supply-chain analysis. - No known CVEs affecting this package version or its production dependencies. - Runs hatchling.build at install time, a recognised build step with no custom scripting around it. - 5 of 40 dependencies flagged as unhealthy. - **Provenance & Transparency**: 100/100 - Source repository is publicly reachable at the declared URL. - Cryptographically verified build provenance (signed, bound to MohammadHijjawi97/since-cutoff). - Clear OSI-approved license (MIT). - Actively maintained (last published 3 days ago). - Publishes a security disclosure policy (SECURITY.md). - **Schema Quality & AI Usability**: 61/100 - AI-judged instruction clarity (excellent). - Context-footprint check failed: tool/resource definitions use about 2097 tokens (~699/item across 3 items; 3 tools + 0 resources), over budget; trim descriptions and params. - Usage-examples check failed: none of the tools include examples. - **Stability & Change Management**: 17/100 - Stability observed for 5 of 30 days with no destabilising changes; credit accrues until the full window elapses. - **Tool Coverage**: 100/100 - 100% of tools have a non-trivial description (not blank, and not just the tool's name). - 100% of tool parameters carry a description. - **Tool Safety**: 100/100 - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured. - We read all 3 captured tool definition(s), and no name or description among them implies an irreversible operation. - An AI judge read all 4 captured unit(s) of tool text and found none that tries to manipulate the model reading it. - **Capabilities**: 100/100 - Implements a current MCP spec version (2026-07-28). ## Install ### How do I install the since-cutoff MCP server? since-cutoff runs locally as a PyPI package, launched with uvx since-cutoff. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation. ### Claude ```bash claude mcp add mohammadhijjawi97-since-cutoff -- uvx since-cutoff ``` ### Cursor ```json { "mcpServers": { "mohammadhijjawi97-since-cutoff": { "command": "uvx", "args": [ "since-cutoff" ] } } } ``` ### VS Code ```json { "servers": { "mohammadhijjawi97-since-cutoff": { "command": "uvx", "args": [ "since-cutoff" ] } } } ``` ### Codex ```bash codex mcp add mohammadhijjawi97-since-cutoff -- uvx since-cutoff ``` ### opencode ```json { "$schema": "https://opencode.ai/config.json", "mcp": { "mohammadhijjawi97-since-cutoff": { "type": "local", "command": [ "uvx", "since-cutoff" ], "enabled": true } } } ``` ### OpenClaw ```bash openclaw mcp add mohammadhijjawi97-since-cutoff --command uvx --arg since-cutoff ``` ### Hermes ```yaml mcp_servers: mohammadhijjawi97-since-cutoff: command: "uvx" args: ["since-cutoff"] ``` ### Netclaw ```json { "McpServers": { "mohammadhijjawi97-since-cutoff": { "Transport": "stdio", "Command": "uvx", "Arguments": [ "since-cutoff" ] } } } ``` ### Vellum ```bash assistant mcp add mohammadhijjawi97-since-cutoff -t stdio -c uvx -a since-cutoff ``` ### Other ```json { "mcpServers": { "mohammadhijjawi97-since-cutoff": { "command": "uvx", "args": [ "since-cutoff" ] } } } ``` ## Changelog Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims. ### 2026-10-01 (score 81, +1) No change was recorded against any check on this day. Stability & Change Management went from 10 to 13. That category is still filling its 30-day observation window: 3 days of observed history at the previous scan, 4 at this one. The score rises as the window fills, whether or not the server changes. ### 2026-09-29 (score 80, +16) - [security improvement] Malware scan: unverified → pass ### 2026-09-28 (score 64, −5) - [security regression] Malware scan: pass → unverified - [security regression] Tool safety: pass → unverified - [security improvement] Provenance: fail → pass - [security] Stability: Stability not yet verified: we do not have a sandbox capture of the MCP schema this version of the package serves yet. - [security] The attested source repository moved: MohammadHijjawi97/since-cutoff - [functional regression] Schema quality: 609 → 699 - [functional regression] Schema quality: 609 → 676 - [functional regression] Capabilities: pass → unverified - [functional regression] Tool coverage: 100 → unverified - [functional improvement] Stability: unverified → 0.03 - [functional] First check of Schema quality: unverified - [functional] Package version: 0.3.2 → 0.5.0 - [functional] Package version: 0.3.2 → 0.4.1 - [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server ### 2026-09-27 (score 69) First indexed and scored. ## MCP tools (3) ### `model_cutoff` (~284 tokens) Model training cutoff Look up a model's training cutoff: library releases after it may be missing from its training data. Use it to find or confirm your own cutoff. The other two tools do not need it first: they take `model` and look the cutoff up themselves. Read-only and quick (about a second): no model is called. The cutoffs come from the models.dev catalogue, fetched at most once a day and cached; without network access a cached or bundled copy is used. Returns one short paragraph of plain text: " (, ): training cutoff YYYY-MM-DD. Released YYYY-MM-DD. Source: models.dev", then the api_changes and project_changes calls to make with that id. An unknown id is a tool error that lists close matches. Input parameters: - `model` (string, required): a model id, e.g. "claude-haiku-4-5", "claude-sonnet-4-5", "claude-opus-4-6[1m]", "gpt-5.4", "gemini-2.5-pro", "anthropic/claude-haiku-4.5" or "openai:gpt-5.4". Pass your own id. The aliases "sonnet",… ### `api_changes` (~727 tokens) Library API changes since the cutoff List the public API changes of one PyPI package since a model's training cutoff. Use it before writing code against a library version that may have been released after your training cutoff (check the lockfile or requirements for the version), or when code fails on a name, import or parameter of that library. Pass `symbol` to see only the functions or classes you are about to use. For all dependencies of a project in one call, use project_changes. Compares the newest final release on or before the cutoff (or `from_version`) with \`to_version` (default: the latest final release on PyPI). Read-only: downloads the two releases' wheels from PyPI (80 MB at most each, by default) and reads their sources statically. No package code runs and no model is called. The first call for a package takes a few seconds, up to a minute or two for very large packages; the result is cached, and later calls take about a second. One of `model`, `cutoff` or `from_version` is required. Returns Markdown: a "# -> " heading; bullets with both versions and their release dates and the number of breaking changes and new deprecations by kind; then sections "Dependencies switched" (the package requires another library instead of one it required, such as `httpx2` instead of `httpx`, and its signatures take that library's objects), "Removed or moved", "Parameters removed", "Parameters now required", "Changed kind", "Now keyword-only or positional-only" and "Deprecated", one line per change with what to use instead when the diff knows it (the new import, the new signature, a parameter probably renamed in place, what the old version's deprecation text said), and names that merely look similar, labelled as not confirmed. A change reachable under several import paths is listed once. If the package had no release by the cutoff, a short note says its whole API was released after your reported training cutoff. Input parameters: - `cutoff`: your training cutoff instead of `model`, as "YYYY-MM" or "YYYY-MM-DD", e.g. "2025-02". - `from_version`: compare from this version instead of the one at the cutoff, e.g. "0.29.1". - `limit` (integer): the most changes to list, e.g. 100 (default 40), shared between the kinds of change. - `model`: your model id, e.g. "claude-haiku-4-5" or "gpt-5.4"; its training cutoff is looked up. - `package` (string, required): the PyPI name, e.g. "huggingface-hub" or "openai", optionally pinned: "anthropic==1.8.0" (the pin is used as `to_version`). - `symbol`: only changes to what this names, e.g. "hf_hub_download", "Messages.create", "client.messages.create" or "client.chat.completions.create". Call syntax and variable names are ignored: the last one or t… - `to_version`: the version the project uses, e.g. "2.0.0" (default: the latest release on PyPI). ### `project_changes` (~752 tokens) Project dependency changes since the cutoff Check every dependency of a Python project for API changes since a training cutoff. Use it at the start of work on a project, or before adding code that uses its dependencies: one call covers every direct dependency at the version the project pins. For a single package, a single symbol, or a version the project does not pin, api_changes is faster. Reads the project's lockfile (uv.lock, poetry.lock, pdm.lock, pylock.toml, Pipfile.lock), requirements*.txt, pyproject.toml or .venv, and for each direct dependency compares the release that existed at the cutoff with the pinned one. Changes to names the project's code uses, in packages it imports, come first. Read-only: nothing in the project is written. Reads PyPI metadata, downloads the wheels of the dependencies that changed and reads them statically; no package code runs and no model is called. The first call on a project with many dependencies can take several minutes (large packages such as transformers take longest); results are cached, so later calls take seconds. Running `since-cutoff scan` in the project once fills the same cache. Reports progress while it works if the client asks for it. One of `model` or `cutoff` is required. Returns Markdown: a heading with the number of dependencies checked and where their versions came from; bullets with the cutoff, how many of the changed APIs the project's code uses, and how many dependencies changed, were first released after the cutoff, are unchanged, or could not be checked; then "## Your code uses these changed APIs": each changed API the code uses, with what changed, "old form" (the code uses it as the release at the cutoff allowed) or "uses this API", the files that use it (at most 3), the note to follow with its tag ([diff]: from the static diff; [library]: the replacement is named in the library's own deprecation text), the runtime caveat when the pinned source still accepts a removed name, and similar names, not confirmed as replacements; then one "##