v0.1.4

Fixed

• HTML comments were printed into the PDF instead of being dropped. With
  markdown-it's html: false, comments are escaped into visible text rather than
  ignored, so a file carrying <!-- markcv:no-build --> rendered that marker at
  the top of the page. Comments are metadata for tooling, never content, and are
  now stripped before rendering.

v0.1.3

Changed

• Package description and keywords rewritten around what people actually search
  for. resume-linter had no package competing for it at all, while markdown resume is crowded — so the wording leads with the linting and page-fit work
  rather than with rendering, which is the part everyone already does.
• README opens by stating what the tool does differently, and adds a section
  comparing it with the existing Markdown resume renderers, including when to use
  one of those instead.
• Image alt text now describes the resumes shown rather than repeating the tool name.

v0.1.2

Fixed

• markcv --version reported the version that was hardcoded in the source rather
  than the one actually installed: 0.1.1 shipped while --version still said
  0.1.0. Both the CLI and the MCP server now read it from package.json, and a
  test fails if the two ever disagree.

v0.1.1

Added

• markcv mcp subcommand, so MCP clients can use the shape everyone else uses:
  npx -y @phuthuycoding/markcv mcp. With two binaries in the package, the bare
  npx form would otherwise need --package=…, which nobody writes from memory.

Changed

• README now leads with installing from npm, and no longer asks the reader to
  clone and build. Development setup moved to CONTRIBUTING.md.
• MCP documentation defaults to the npx form; the global markcv-mcp binary is
  presented as the faster alternative rather than the only way.
• render --no-photo reports photo disabled instead of
  photo none (placeholder box) — not embedding a photo and not finding one are
  different things.

v0.1.0 — first release

First release of fitcv.

Build a CV from Markdown, as a CLI and an MCP server for AI agents.

What it does that a plain markdown-to-PDF tool does not

• fit separates content too long from bad page break, and names the block that got pushed onto a new page. Trimming words is the wrong fix when a heading is simply sitting too close to a page boundary — that one needs a reorder.
• lint audits content rather than formatting, catching both overselling (spearheaded, comprehensive) and underselling (advised on next to work of real scale).

Also included

• tailor compares a CV against a job description: what is missing, what is claimed only under SKILLS, which bullets are unrelated.
• Master/variant workflow, portrait photo auto-detection, two ATS-safe themes.
• MCP server with 7 tools returning structured JSON, so an agent can iterate on its own.

Uses puppeteer-core against a system Chrome — no bundled Chromium.