Skip to content

deps: bump beam_mcp from 0.8.0 to 0.10.1 - #14

Merged
HackTuah merged 1 commit into
mainfrom
dependabot/hex/beam_mcp-0.10.1
Oct 7, 2026
Merged

HackTuah merged 1 commit into
mainfrom
dependabot/hex/beam_mcp-0.10.1

Conversation

@dependabot

@dependabot dependabot Bot commented on behalf of github Sep 30, 2026 •

Copy link
Copy Markdown
Contributor

Bumps beam_mcp from 0.8.0 to 0.10.1.

Changelog

Sourced from beam_mcp's changelog.

[0.10.1] - 2026-09-26

A security release. Found by the project's own security review (three independent review lanes, their findings reproduced before any fix). No public entry added, removed, renamed, hidden or changed in arity (docs/public-api.txt did not move). Two behaviours change because the old ones were the defects; each entry says how to tell whether you are affected. Take the patch: ~> 0.10.0 already admits it. beam_mcp_signer 0.2.1 is released beside it.

Security: JSON true, false and null reach dispatch as themselves

  • A tool argument of JSON false reached dispatch as the string "false", true as "true" and null as "nil", at any depth; a string is truthy, so a host testing if args.dry_run ran the non-dry path on "dry_run": false. They now arrive as false, true and nil, and a host result carrying them goes out as JSON booleans and null. How to tell whether you are affected: a tool that takes a boolean or a nullable argument. A host that compared against the strings (args.flag == "true") must compare against the booleans.

Security: the advertised schema is enforced at every depth; a keyword it would not enforce is refused

  • BeamMCP.Schema checked type, required and additionalProperties at the top level only; nested objects, enum, items, maxItems, maxLength, pattern and the rest were advertised by tools/list and not held at tools/call. Every keyword of the subset the README lists is now enforced at every depth, a refusal naming the path (opts.mode, tags[1]). A keyword outside the subset (oneOf, anyOf, $ref, if, not and the rest) is refused when the catalog is loaded (BeamMCP.Server.new/1 raises, naming the tool and the keyword) rather than advertised and ignored; annotations (title, description, default, examples, format and the like, and x- keys) remain allowed. How to tell whether you are affected: a tool whose input_schema nests objects or uses any keyword beyond type, properties, required and additionalProperties. A catalog using an unsupported keyword now fails at startup: express the constraint in the subset, or check it in the host's dispatch.
  • How the subset reads, stated in BeamMCP.Schema's moduledoc: pattern uses ECMA-262's character classes and anchors (\d, \w, \s ASCII; $ never before a trailing newline), so ^[a-z]+$ refuses "abc\n"; on OTP 27 only, \w, \b and the POSIX letter classes also match the Latin-1 letters U+00AA to U+00FF, since that release's :re has Latin-1 tables and no ASCII ones; enum, const and uniqueItems compare as JSON (1 equals 1.0); integer refuses 1.0, stricter than JSON Schema, so a host typed for integers never receives a float. A property name that is not a string, and an enum or const value that is not JSON, are refused when the catalog is loaded.

Security: on stdio, nothing but protocol messages reaches standard output

  • The stdio transport logged a host fault through Logger, whose default console handler writes to standard output, the protocol's stream: the report (a stacktrace, a raise's message) reached the client as a line of the protocol. It now writes the report, with arities in place of arguments, to standard error. A response the encoder refuses (a host term that is not valid JSON) is answered -32603 with its id, and the loop goes on where it had ended. The README now says to route a stdio host's own Logger to standard error.

... (truncated)

Commits
  • 04548e1 pattern on OTP 27: \w, \b and the POSIX letter classes also take the Latin-1 ...
  • 289dbdb Release 0.10.1: a security patch from the project's own review; no public ent...
  • 81958da The bus factor is two: znmead knows the code and reviews pull requests
  • 0b2a63a The slice records leave the public tree; SLICES.md says what each slice commi...
  • cc3680f ci: bump github/codeql-action/upload-sarif from 4.38.0 to 4.38.1
  • 027fcc9 Release 0.10.0: the quiet minor again (FIPS in CI, an attested SBOM at releas...
  • f570555 SBOM at release: CycloneDX, generated outside mix.exs by the EEF's tool, atte...
  • 4176120 FIPS leg: the suite in FIPS mode in CI, on the validated 3.1.2 provider; docs...
  • d81228f README's Scheduled paragraph follows the road: federation and effective conne...
  • 8a16009 The em-dash sweep (CX-141): none left outside slices/, the CHANGELOG's releas...
  • Additional commits viewable in compare view

@dependabot dependabot Bot added dependencies Pull requests that update a dependency file elixir Pull requests that update elixir code labels Sep 30, 2026
@HackTuah

HackTuah commented Oct 7, 2026

Copy link
Copy Markdown
Member

@dependabot rebase

@dependabot
dependabot Bot force-pushed the dependabot/hex/beam_mcp-0.10.1 branch from 3c5b8af to 39e13b7 Compare October 7, 2026 23:41
Bumps [beam_mcp](https://github.com/ScriptKittyOS/beam_mcp) from 0.8.0 to 0.10.1.
- [Changelog](https://github.com/ScriptKittyOS/beam_mcp/blob/main/CHANGELOG.md)
- [Commits](ScriptKittyOS/beam_mcp@v0.8.0...v0.10.1)

---
updated-dependencies:
- dependency-name: beam_mcp
  dependency-version: 0.10.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
@dependabot
dependabot Bot force-pushed the dependabot/hex/beam_mcp-0.10.1 branch from 39e13b7 to 9a442c1 Compare October 7, 2026 23:43
@HackTuah
HackTuah merged commit ce5d178 into main Oct 7, 2026
5 checks passed
@dependabot
dependabot Bot deleted the dependabot/hex/beam_mcp-0.10.1 branch October 7, 2026 23:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dependencies Pull requests that update a dependency file elixir Pull requests that update elixir code

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant