Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: contentstack/launch-cli
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: main
Choose a base ref
...
head repository: contentstack/launch-cli
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: development
Choose a head ref
Checking mergeability… Don’t worry, you can still create the pull request.
  • 4 commits
  • 12 files changed
  • 3 contributors

Commits on Nov 5, 2026

  1. fix(deps): refresh rimraf to 6.1.4 to clear the glob advisory on the …

    …nyc > rimraf path
    
    Semver-compatible lockfile refresh (npm update rimraf, npm 11): rimraf
    6.1.3 -> 6.1.4 moves its glob from 13.0.6 to 14.0.3, clearing
    GHSA-3g3m-2fjr-6pw4 on the nyc > rimraf path. Dev-only. A full
    `npm audit fix` was tried first and reverted: it also raises @oclif/core
    to 4.14.0, which breaks 30/31 integration tests (ESM-only wsl-utils).
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    Claude-Session: https://claude.ai/code/session_01NTFwVv9F3TgYAwMxADEnGj
    claude committed Nov 5, 2026
    Configuration menu
    Copy the full SHA
    f605f3a View commit details
    Browse the repository at this point in the history

Commits on Nov 7, 2026

  1. Merge pull request #284 from contentstack/CL-snyk-fixes/05-Nov-2026

    fix(deps): refresh rimraf to 6.1.4 to clear the glob advisory on the nyc > rimraf path (+Claude)
    dhruv-parekh-cs authored Nov 7, 2026
    Configuration menu
    Copy the full SHA
    1e10888 View commit details
    Browse the repository at this point in the history

Commits on Nov 17, 2026

  1. fix(deps): upgrade open to 11 and @oclif/core to 4.14 to clear Snyk f…

    …indings (#286)
    
    * style: fix existing eslint errors in base-class.ts and open.ts
    
    The next commit changes both files, and lint-staged runs eslint --fix on
    every staged file, so their existing errors would block it.
    
    eslint --fix: indentation, quote style, let to const, Array<T> to T[].
    By hand: drop an unused catch parameter, wrap two JSDoc lines, and add
    eslint-disable-next-line for long user-facing strings, one URL and the
    require() of the user's config file, so none of that text or behaviour
    changes. No functional change.
    
    .talismanrc gets a checksum entry for base-class.ts: Talisman flags the
    moved "key" type annotations (parseEnvVariablesString, envKeys) as
    potential secrets; they are not.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix(deps): upgrade open to 11 and @oclif/core to 4.14 to clear Snyk findings
    
    open@8.4.2 (SNYK-JS-OPEN-22112123, High) is fixed only in open 11, which
    is ESM-only. The CLI compiles to CommonJS, where TypeScript turns import()
    into require(), so the three browser-open call sites now go through
    util/open-url.ts, which loads open with a native dynamic import. The
    GitHub-connect step and launch:open await it, so the process cannot exit
    before the browser starts; the delayed deployment-URL open keeps its
    fire-and-forget timer. Opening the browser stays best effort: open 11
    rejects when the launcher fails, and every caller prints the URL first,
    so openUrl swallows that instead of failing the command (or crashing
    with an unhandled rejection after a successful deploy).
    
    @oclif/core moves from 4.11.14 to 4.14.0 within the existing ^4.2.7 range
    (lockfile only). 4.14 uses ejs 6, which drops the ejs > jake > filelist >
    minimatch@5.1.9 path behind SNYK-JS-MINIMATCH-22111072. An override pins
    @oclif/core's own minimatch to ^11.0.1, as on v2-dev.
    
    Two findings remain reported, not ignored. Neither has an upgrade, so
    neither fails --fail-on=all: adm-zip@0.6.1 (SNYK-JS-ADMZIP-21707815), and
    open@8.4.2 pulled in by @contentstack/cli-utilities, whose latest release
    (2.0.2) still depends on open ^8.4.2.
    
    Integration tests now mock util/open-url instead of open, and stub
    @oclif/core's getPlatform, which 4.14 implements with a native
    import('wsl-utils') that Jest's VM cannot run.
    
    .talismanrc: base-class.ts checksum refreshed for its new content.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * test: keep the integration harness's git calls out of the repo when run from a hook
    
    Git exports GIT_DIR and the other repo-local variables to hooks, and the
    pre-push hook runs npm test. With them set, the harness's git init,
    config, commit and remote add in its temp projects acted on this
    repository: a push from a worktree left "initial commit" commits on the
    branch and set user.name, user.email, commit.gpgsign and core.bare in the
    shared .git/config.
    
    jest.global-setup.ts now clears the variables `git rev-parse
    --local-env-vars` lists. It runs in Jest's parent process, so the child
    processes the harness and the CLI spawn inherit the clean environment
    (setupFiles only change the sandbox's copy of process.env).
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix: address review on the open helper, lockfile and harness docs
    
    - open-url.ts: create the dynamic importer lazily inside openUrl's try.
      Built at module load, new Function threw EvalError under
      --disallow-code-generation-from-strings, so every command importing the
      helper crashed instead of just not opening a browser.
    - openUrl takes an optional loader, and the tests inject a fake open to
      cover the success path (called with the URL) plus both failure paths,
      instead of relying on Jest's VM rejecting the native import.
    - Drop the code comments added in this PR (AGENTS.md: no code comments);
      the rationale lives in the commit messages and the PR description.
    - AGENTS.md: the integration harness now replaces src/util/open-url and
      @oclif/core's getPlatform rather than the open package, and
      jest.global-setup.ts clears git's repo-local env vars.
    - @oclif/core range raised to ^4.14.0 so installs cannot keep a 4.11.x
      that still carries the flagged minimatch path.
    - package-lock.json: restore the 23 "libc" fields npm 10 dropped, copied
      from development for the same package versions. npm 11 (release CI)
      re-resolves the file without changes.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix: address second review on the harness docs and openUrl tests
    
    - AGENTS.md: the "all run for real" sentence had been joined onto the
      new git-env bullet; it closes the "Only these are replaced" bullet
      again, so the harness contract reads as before.
    - open-url.test.ts: the failure-path tests now keep the happy path's
      assertions (AGENTS.md: do not skip assertions in edge-case tests). The
      load-failure test checks the loader was called; the open-failure test
      checks open was called with the URL.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    ---------
    
    Co-authored-by: Claude <noreply@anthropic.com>
    Harshi-Shah-CS and claude authored Nov 17, 2026
    Configuration menu
    Copy the full SHA
    080cd70 View commit details
    Browse the repository at this point in the history
  2. fix(deps): declare tslib as a runtime dependency (#287)

    * style: fix existing eslint errors in base-class.ts and open.ts
    
    The next commit changes both files, and lint-staged runs eslint --fix on
    every staged file, so their existing errors would block it.
    
    eslint --fix: indentation, quote style, let to const, Array<T> to T[].
    By hand: drop an unused catch parameter, wrap two JSDoc lines, and add
    eslint-disable-next-line for long user-facing strings, one URL and the
    require() of the user's config file, so none of that text or behaviour
    changes. No functional change.
    
    .talismanrc gets a checksum entry for base-class.ts: Talisman flags the
    moved "key" type annotations (parseEnvVariablesString, envKeys) as
    potential secrets; they are not.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix(deps): upgrade open to 11 and @oclif/core to 4.14 to clear Snyk findings
    
    open@8.4.2 (SNYK-JS-OPEN-22112123, High) is fixed only in open 11, which
    is ESM-only. The CLI compiles to CommonJS, where TypeScript turns import()
    into require(), so the three browser-open call sites now go through
    util/open-url.ts, which loads open with a native dynamic import. The
    GitHub-connect step and launch:open await it, so the process cannot exit
    before the browser starts; the delayed deployment-URL open keeps its
    fire-and-forget timer. Opening the browser stays best effort: open 11
    rejects when the launcher fails, and every caller prints the URL first,
    so openUrl swallows that instead of failing the command (or crashing
    with an unhandled rejection after a successful deploy).
    
    @oclif/core moves from 4.11.14 to 4.14.0 within the existing ^4.2.7 range
    (lockfile only). 4.14 uses ejs 6, which drops the ejs > jake > filelist >
    minimatch@5.1.9 path behind SNYK-JS-MINIMATCH-22111072. An override pins
    @oclif/core's own minimatch to ^11.0.1, as on v2-dev.
    
    Two findings remain reported, not ignored. Neither has an upgrade, so
    neither fails --fail-on=all: adm-zip@0.6.1 (SNYK-JS-ADMZIP-21707815), and
    open@8.4.2 pulled in by @contentstack/cli-utilities, whose latest release
    (2.0.2) still depends on open ^8.4.2.
    
    Integration tests now mock util/open-url instead of open, and stub
    @oclif/core's getPlatform, which 4.14 implements with a native
    import('wsl-utils') that Jest's VM cannot run.
    
    .talismanrc: base-class.ts checksum refreshed for its new content.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * test: keep the integration harness's git calls out of the repo when run from a hook
    
    Git exports GIT_DIR and the other repo-local variables to hooks, and the
    pre-push hook runs npm test. With them set, the harness's git init,
    config, commit and remote add in its temp projects acted on this
    repository: a push from a worktree left "initial commit" commits on the
    branch and set user.name, user.email, commit.gpgsign and core.bare in the
    shared .git/config.
    
    jest.global-setup.ts now clears the variables `git rev-parse
    --local-env-vars` lists. It runs in Jest's parent process, so the child
    processes the harness and the CLI spawn inherit the clean environment
    (setupFiles only change the sandbox's copy of process.env).
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix(deps): declare tslib as a runtime dependency
    
    tsconfig sets importHelpers, so the compiled commands require tslib at
    runtime, but it was declared only as a devDependency. When csdx installs
    @contentstack/cli-launch as a plugin it skips devDependencies, so a strict
    install (pnpm) fails every launch command with "Cannot find module
    'tslib'" (#194). Default npm installs only survived because tslib was
    hoisted from another package.
    
    Also drop @oclif/plugin-plugins from oclif.plugins. Its dependency was
    removed in February, so oclif already skipped the entry; help output and
    the generated manifest are unchanged without it.
    
    Add the dependency guard test from v2-dev (dc69e11). It compiles every
    runtime source with the project's compiler options and fails on any
    required package, or oclif.plugins entry, that is not a production
    dependency.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix: address review on the open helper, lockfile and harness docs
    
    - open-url.ts: create the dynamic importer lazily inside openUrl's try.
      Built at module load, new Function threw EvalError under
      --disallow-code-generation-from-strings, so every command importing the
      helper crashed instead of just not opening a browser.
    - openUrl takes an optional loader, and the tests inject a fake open to
      cover the success path (called with the URL) plus both failure paths,
      instead of relying on Jest's VM rejecting the native import.
    - Drop the code comments added in this PR (AGENTS.md: no code comments);
      the rationale lives in the commit messages and the PR description.
    - AGENTS.md: the integration harness now replaces src/util/open-url and
      @oclif/core's getPlatform rather than the open package, and
      jest.global-setup.ts clears git's repo-local env vars.
    - @oclif/core range raised to ^4.14.0 so installs cannot keep a 4.11.x
      that still carries the flagged minimatch path.
    - package-lock.json: restore the 23 "libc" fields npm 10 dropped, copied
      from development for the same package versions. npm 11 (release CI)
      re-resolves the file without changes.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * test: catch single-quoted and string-built imports in the dependency guard
    
    The guard only matched double-quoted require("x"). TypeScript emits a
    dynamic import('x') as require('x') with single quotes, and the open
    helper builds import("open") inside a string passed to new Function, so
    an undeclared package pulled in either way passed the guard and then
    failed at runtime. It now matches require(...) and import(...) with
    either quote style, and asserts that open is detected.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    * fix: address second review on the harness docs and openUrl tests
    
    - AGENTS.md: the "all run for real" sentence had been joined onto the
      new git-env bullet; it closes the "Only these are replaced" bullet
      again, so the harness contract reads as before.
    - open-url.test.ts: the failure-path tests now keep the happy path's
      assertions (AGENTS.md: do not skip assertions in edge-case tests). The
      load-failure test checks the loader was called; the open-failure test
      checks open was called with the URL.
    
    #claude_code#
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-Authored-By: Claude <noreply@anthropic.com>
    
    ---------
    
    Co-authored-by: Claude <noreply@anthropic.com>
    Harshi-Shah-CS and claude authored Nov 17, 2026
    Configuration menu
    Copy the full SHA
    f6f7503 View commit details
    Browse the repository at this point in the history
Loading
⌂ Proxy