agent-smoke-e2e
End-to-end smoke check of the moraine MCP server. Exercises the public retrieval workflow against a live moraine stack and prints…
Start Moraine repository development work safely. Use when an agent takes on new code, docs, test, CI, packaging, or release work in this repo; when it must create or verify an isolated branch/worktree; when sibling agent collisions are possible; or before editing files for a
$ npx -y skills add eric-tramel/moraine --skill moraine-start-work --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/moraine-start-workContext preview
The summary Claude sees to decide when to auto-load this skill.
Start Moraine repository development work safely. Use when an agent takes on new code, docs, test, CI, packaging, or release work in this repo; when it must create or verify an isolated branch/worktree; when sibling agent collisions are possible; or before editing files for a
name: moraine-start-work description: Start Moraine repository development work safely. Use when an agent takes on new code, docs, test, CI, packaging, or release work in this repo; when it must create or verify an isolated branch/worktree; when sibling agent collisions are possible; or before editing files for a Moraine contributor task.
Use this workflow to start a Moraine contributor task with the same orientation every time: verify the checkout, isolate the branch, read governing instructions, and choose the right validation path before editing.
Using `$moraine-dev:code-review` in Codex or `/code-review` in Kiro is an explicit user request for delegated reviewer subagents. You have permission to launch that review wave when the follow-on workflow reaches that step.
Use the invocation form native to the current harness:
| Workflow | Codex | Kiro | | --- | --- | --- | | `crystallize` | `$moraine-dev:crystallize` | `/crystallize` | | `code-review` | `$moraine-dev:code-review` | `/code-review` | | `moraine-sandbox-qa` | `$moraine-dev:moraine-sandbox-qa` | `/moraine-sandbox-qa` | | `moraine-author-pr` | `$moraine-dev:moraine-author-pr` | `/moraine-author-pr` |
After the start checklist and branch/worktree setup, run the rest of the contributor workflow in this order:
1. Use the current harness's `crystallize` invocation to turn rough or ambiguous input into a ready-to-implement local plan. 2. Implement the scoped change from that plan or, when the task is already concrete, from the user's explicit request. 3. Use the current harness's `code-review` invocation and iterate fixes until the review is clean or every remaining item is explicitly rejected or deferred with rationale. 4. Use the current harness's `moraine-sandbox-qa` invocation for stack-facing changes and iterate fixes until sandbox QA is clean. 5. Use the current harness's `moraine-author-pr` invocation to create a PR.
Run these from the repository root unless the current harness already supplied a trusted workspace root:
pwd git rev-parse --show-toplevel git status --short --branch git worktree list --porcelain find .. -name AGENTS.md -print
If the user refers to previous work, another agent, an unexplained decision, a branch you cannot see, or a failure from another session, use the Moraine MCP tools directly before making assumptions.
Use an isolated branch for development work. Choose a repository-appropriate prefix such as `feat/`, `fix/`, or `docs/`; branch names describe the work, not the harness running it.
If the harness already placed you in an isolated worktree, create or verify a branch there:
branch="feat/<short-task-name>" # replace with the repository-appropriate prefix git switch -c "$branch"
If you are in the main checkout and need to start a new worktree, create one with a task-specific name:
branch="feat/<short-task-name>" # replace with the repository-appropriate prefix git worktree add ../moraine-worktrees/<short-task-name> -b "$branch" main
When spawned into a temporary agent worktree, compare your `HEAD` with any sibling worktree whose branch name matches the task. If the sibling has newer in-progress code, treat that sibling as authoritative and run sandbox validation from that worktree.
Favor narrative description of what you tested instead of bullet points.
Moraine is a local trace stack for agent work. It indexes sessions from agent harnesses such as Codex, Claude Code, Qwen Code, Kiro CLI, Kimi CLI, NAC, OpenCode, Hermes, Pi Coding Agent, and Prime Agent into ClickHouse, serves a monitor UI, and exposes MCP
End-to-end smoke check of the moraine MCP server. Exercises the public retrieval workflow against a live moraine stack and prints…
Rewrite a moraine GitHub release body in the house format — one-sentence pitch, install block, usage-focused "what's new"…
Help prepare and, only with explicit user approval, file a sanitized Moraine bug report for reproducible defects in Moraine…
Use Moraine realtime session data to inspect active agent work across harnesses when the user asks what is happening now or what…
Use Moraine MCP tools to recover prior agent-session context when the user refers to earlier work, decisions, failures, branches,…
Review a PR through the CodeReviewCompleteness persona. Use when asked to check whether a PR fully addresses the objectives it…