Adopt --json on remaining script-facing commands (decisions --list, clean/cleanup --dry-run, validate)

[pbean]

Tracking: adopt --json (pure-document contract) on the remaining script-facing commands

status --json (#190) and list --json (#192) emit stable, schema-versioned pure JSON documents per the contract in src/bmad_loop/machine.py. Other commands whose text output is a de-facto API for scripts are still text-only:

• decisions --list (Adopt --json on decisions --list (pure-document contract) #203) — pending decisions with ids/options; scripts currently scrape the numbered text.
• clean / cleanup --dry-run (Adopt --json on clean and cleanup (pure-document contract) #204) — what would be removed (worktrees, runs past retention, tmux sessions); the natural machine surface for pre-flight checks.
• validate (Adopt --json on validate (pure-document contract) #205) — pass/fail plus per-check findings; exit code alone loses the detail.

Now split into the per-command issues linked above — each needs its own *_SCHEMA_VERSION, document builder, and purity tests via the shared machine_json helper in tests/conftest.py. Not a commitment to do all of them; strike any that turn out to have no script consumers.

Related: #195 tracks unifying the older report+embedded-block form of diagnose --json / probe-adapter --json with this contract.

[coderabbitai]

🔗 Related PRs

#193 - feat(cli): add status --json stable document (#190) [merged]
#197 - refactor(cli): codify the machine-output (--json) contract [merged]
#198 - feat(cli): add list --json stable document [merged]
#201 - refactor(cli): unify diagnose/probe-adapter --json with the pure-document contract [merged]
#206 - chore(tests): share the --json purity helper and de-hardcode the command count [open]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[pbean]

Scope fully shipped — closing. All three per-command issues landed, but none of the PRs used a closing keyword against this tracking issue, so it stayed open.

command	issue	PR
decisions --json	#203	#207 (merged)
clean / cleanup --json	#204	#208 (merged)
validate --json	#205	#209 (merged)

Each shipped its own *_SCHEMA_VERSION (all at 1), a document builder, and purity tests through the shared machine_json helper in tests/conftest.py, as planned above. Nothing was struck — all three had real script consumers.

The successor thread is #210: the TUI's validate modal should consume validate --json and render the findings structurally instead of scraping text. That is the first in-tree consumer of one of these documents, and the stepping stone toward a web UI reading the same contract.

Also relevant to anyone picking that up: #211 extracts the six document builders out of cli.py into a new documents.py library layer, so a non-CLI frontend can import them directly rather than shelling out to the CLI. (#195 remains the separate thread for diagnose / probe-adapter.)