v0.0.9

[0.0.9] - 2026-04-08

New Features

• GitHub Actions annotations — --report github outputs ::error workflow commands with file and line numbers. Failed assertions appear as inline annotations on PR diffs, pointing to the exact Markdown line that broke.

  mdproof --report github runbooks/

  ::error file=runbooks/api-proof.md,line=12::Step 1 (Health check): expected output (not found in stdout)
  

  Combine with JUnit XML for both annotations and test summary:

  - run: mdproof --report github runbooks/
  - run: mdproof --report junit -o results.xml runbooks/

Documentation

• Prerequisites section — README now lists system requirements (Docker for sandbox, jq for jq assertions) before install instructions
• Troubleshooting guide — new docs/troubleshooting.md covers strict mode errors, jq setup, snapshot mismatches, timeouts, and common CLI issues
• CLI discoverability — mdproof --help tip added to Quick Start section

Full Changelog: v0.0.8...v0.0.9

v0.0.8

[0.0.8] - 2026-04-06

New Features

• Working directory — --workdir DIR sets the working directory for all step execution. Supports shell expansion (e.g., $HOME). Also configurable in mdproof.json via workdir. CLI flag overrides config.

  mdproof --workdir /tmp/workspace deploy-proof.md
  mdproof --workdir '$HOME/project' test-proof.md

  { "workdir": "/tmp/workspace" }

Bug Fixes

• Working directory not applied — fixed runFile() not threading Workdir from RunOptions to the executor, which silently ignored --workdir
• Shell quoting for workdir — the generated cd command now uses double quotes, preventing breakage with paths containing spaces and protecting against shell injection while preserving $VAR expansion

Full Changelog: v0.0.7...v0.0.8

v0.0.7

[0.0.7] - 2026-03-31

New Features

• Retained failure artifacts — --keep-failed-artifacts now preserves the failed runbook's executor artifact directory, and also keeps the isolated $HOME / $TMPDIR directory when --isolation per-runbook is active. You can enable it from the CLI or set keep_failed_artifacts in mdproof.json. The plain-text failure summary points directly to the retained script, env, stdout, and stderr paths so you can inspect the exact failed step without re-deriving temp locations.

  mdproof --keep-failed-artifacts runbooks/fixtures/failing-proof.md
  mdproof --isolation per-runbook --keep-failed-artifacts runbooks/fixtures/failing-proof.md

• Failed-step script and env printing — --print-step-script and --print-step-env dump the actual failed step script or captured execution environment to stderr, which makes debugging wrappers, temp files, and execution context much faster while keeping --report json clean on stdout. These defaults can also live in mdproof.json via print_step_script and print_step_env.

  mdproof --print-step-script --print-step-env runbooks/fixtures/failing-proof.md
  mdproof --print-step-script --print-step-env --report json runbooks/fixtures/failing-proof.md

  mdproof --report json runbooks/fixtures/failing-proof.md | jq '.artifact_dir, .steps[0].debug'

Full Changelog: v0.0.6...v0.0.7

v0.0.6

[0.0.6] - 2026-03-13

New Features

• Source-aware failure reporting — failed assertions, command exits, and parser errors now point to the exact Markdown file and line number. Works in plain text, JSON (steps[].source, assertions[].source), and JUnit output.

  FAIL deploy-proof.md:13 Step 1: Assertion failure
  Assertion deploy-proof.md:13 expected output
  Command deploy-proof.md:7-10
  

  mdproof --report json test.md | jq '.steps[0].source.heading.start.line'

Breaking Changes

• --watch flag removed — watch mode has been removed to sharpen product focus on executable runbooks, docs verification, and sandboxed smoke tests. Use an external file watcher (e.g., entr, watchexec) if you need re-run-on-change behavior.

Full Changelog: v0.0.5...v0.0.6

v0.0.5

[0.0.5] - 2026-03-12

New Features

• Per-runbook isolation — --isolation per-runbook gives each runbook a fresh $HOME and $TMPDIR, preventing cross-runbook pollution in directory mode. Also configurable in mdproof.json via isolation.

  mdproof --isolation per-runbook ./tests/

  { "isolation": "per-runbook" }

• Sub-command variable persistence — variables now persist across --- blocks within the same step. Each sub-command saves its environment via EXIT trap, so export VAR=value (or any assignment, since set -a is active) in block 1 is visible in block 2. This reverses the v0.0.4 breaking change — --- blocks are now isolated subshells that still share environment state.

Bug Fixes

• JSON array output for directory mode — --report json stdout now emits a valid JSON array ([{...}, {...}]) when running multiple runbooks, instead of streaming independent objects. Single-file mode still outputs a single JSON object.

  mdproof --report json ./tests/ | jq '.[].summary'   # directory mode
  mdproof --report json test.md  | jq '.summary'      # single file

Full Changelog: v0.0.4...v0.0.5

v0.0.4

[0.0.4] - 2026-03-12

New Features

• Per-step setup/teardown — -step-setup and -step-teardown CLI flags run a command before/after each step. Setup failure marks the step as failed and skips the body; teardown failure is informational only. Also configurable in mdproof.json via step_setup / step_teardown.

  mdproof -step-setup 'rm -rf /tmp/test-*' test.md
  mdproof -step-teardown 'echo cleanup' test.md

  { "step_setup": "reset-db", "step_teardown": "dump-logs" }

• Sub-command granular report — steps with --- separators now execute each block independently in its own subshell. The JSON report includes a sub_commands array with per-sub-command exit_code, stdout, stderr, and command. Plain text and JUnit reporters surface sub-command failure details.

Breaking Changes

• --- separated blocks no longer share shell variables. Previously, multiple code blocks within a step (delimited by ---) ran as a single concatenated command sharing shell variables. They now run in independent subshells. Migration: if a runbook depends on variables flowing across --- blocks, merge those blocks into a single code block (remove the ---). This is the correct approach — --- semantically means "independent verification step."

Full Changelog: v0.0.3...v0.0.4

v0.0.3

[0.0.3] - 2026-03-11

New Features

• JUnit XML output — --report junit produces JUnit XML for native CI test result display (GitHub Actions, GitLab CI, Jenkins)

  mdproof --report junit tests/          # stdout
  mdproof --report junit -o report.xml tests/  # file output

  Full Changelog: v0.0.2...v0.0.3

v0.0.2

Full Changelog: v0.0.1...v0.0.2

v0.0.1

Full Changelog: https://github.com/runkids/mdproof/commits/v0.0.1