Skip to content
Development
Command

/meta-theory-report

Reopen a Meta_Kim governed run report

From plugin
meta-kim
2633 skills9 agents3 commands
Install
$ npx -y skills add KimYx0207/Meta_Kim --agent claude-code

How it fires

How this command gets triggered: by you, by Claude, or both.

  • Fires itselfClaude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/meta-theory-report

Context preview

What this command does when you run it.

Reopen a Meta_Kim governed run report

Command definition

meta-theory-report.md
description: Reopen a Meta_Kim governed run report
argument-hint: [runId|latest]

Open the user-readable report for a governed Meta_Kim run.

Run this from the installed Meta_Kim package root when the command has been rendered by global sync:

node "__META_KIM_PACKAGE_ROOT__/scripts/run-meta-theory-governed-execution.mjs" --read "${ARGUMENTS:-latest}"

If the package-root placeholder is still present and the current project is the Meta_Kim source checkout, use the source fallback:

npm run meta:theory:report -- "${ARGUMENTS:-latest}"

Report the returned status, runId, markdown report path, and `selection` explanation. `selectionSource` is `latest_committed_pointer` or `explicit_run_id`. `selectionReason` is `last_committed_governed_report` or `explicit_committed_run_id`. Both are stable machine-readable enums. `selectionExplanation` carries the human-readable explanation. In plain language:

  • `selectionSource=latest_committed_pointer` means the newest report that finished committing;

it does not mean that report is the current Claude Code host run.

  • `selectionSource=explicit_run_id` means the requested historical/committed run was opened

directly.

  • `artifactClaimStatus` is the selected artifact's own claim. A valid `partial`

report is still readable and must not be hidden or turned into an error.

  • `different_active_run` means canonical repository lifecycle state identifies

another active run. Say that clearly, then relay `nextCommand` to inspect or continue it. Do not invent a report for it.

  • `unknown_invalid_projection` means repository lifecycle state could not be

validated; a weak `active-run.json` projection is not current-run authority.

  • `not_checked_custom_output` means a custom, temporary, or external report

directory was read without consulting repository lifecycle state.

Never relay task text or task fingerprints from lifecycle state. Do not claim the run is verified unless the report itself contains fresh verification evidence. `meta:run-status --latest --details` and `--json` are redacted status surfaces; use this explicit report readback when the user asks to inspect report content.

Read more
Ships withmeta-kim

Governed execution layer for AI coding assistants: clarify intent, route capabilities, review evidence, verify results, and write back lessons across Claude Code, Codex, OpenClaw, and Cursor.

Get the whole plugin