feat: Add orchestrator support — /ce:run, sync, and /lfg refactor

Orchestrators are pluggable workflow definitions that control which
phases to execute, which reviewers to prioritize, and how to
synthesize findings. They use the same .md format as reviewers
(YAML frontmatter + prose body).

- Add /ce:run skill to load and execute named orchestrators
- Add orchestrator-registry.yaml as default source config
- Generalize sync-reviewers.sh to handle both reviewers and
  orchestrators via a config-name parameter
- Update /ce:refresh to sync orchestrators alongside reviewers
- Refactor /lfg to delegate to /ce:run lfg

Orchestrator format supports:
- phases with gates, skip conditions, and per-phase model overrides
- review-preferences (required/preferred categories, min reviewers)
- synthesis lens (how to weigh findings)
- orchestrator-model vs agent-model for cost control
- personality prose in the markdown body

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: jcasimir

plugins/compound-engineering/skills/ce-run/SKILL.md

@@ -0,0 +1,113 @@
+---
+name: ce:run
+description: "Run a named orchestrator to manage a full engineering workflow. Orchestrators define which phases to execute, which reviewers to prioritize, and how to synthesize findings. Use when you want a specific workflow style — e.g., /ce:run erin for full-process PM, /ce:run max for a quick spike."
+argument-hint: "<orchestrator-name> [feature description]"
+---
+
+# Run Orchestrator
+
+Loads a named orchestrator definition and executes its workflow.
+
+## Step 1: Parse arguments
+
+Split `$ARGUMENTS` into:
+- **Orchestrator name** — the first word (e.g., `erin`, `max`, `lfg`)
+- **Feature description** — everything after the first word
+
+If no orchestrator name is provided, list available orchestrators and ask which to use.
+
+## Step 2: Locate plugin
+
+Find the plugin's install location:
+
+```bash
+PLUGIN_DIR=$(find "$HOME/.claude" "$HOME/.claude-"* -path "*/compound-engineering/*/orchestrators" -type d 2>/dev/null | head -1 | sed 's|/orchestrators$||')
+```
+
+Fall back to relative path if not found:
+
+```bash
+PLUGIN_DIR="${PLUGIN_DIR:-plugins/compound-engineering}"
+```
+
+## Step 3: Load orchestrator
+
+Look for `$PLUGIN_DIR/orchestrators/<name>.md`. If not found:
+
+1. List available orchestrators: `ls $PLUGIN_DIR/orchestrators/*.md`
+2. Show the user what's available
+3. Suggest running `/ce:refresh` if no orchestrators are found
+
+Read the orchestrator file. Parse the YAML frontmatter for structured data (phases, review-preferences, synthesis) and the markdown body for personality/behavior prose.
+
+## Step 4: Adopt the orchestrator persona
+
+Before executing any phases, adopt the orchestrator's personality from the markdown body. This shapes how you communicate, make judgment calls, and interact with the user throughout the workflow.
+
+If the orchestrator has `skip-when` conditions on optional phases, evaluate them against the feature description and current context to decide which phases to include.
+
+## Step 5: Execute phases
+
+For each phase in the `phases` list from frontmatter:
+
+1. **Check if optional** — If the phase has `optional: true` and `skip-when`, evaluate whether to skip based on the feature description and context. Explain your reasoning to the user.
+
+2. **Invoke the skill** — Run the skill specified in `skill:`, passing `args:` with variable substitution:
+   - `$ARGUMENTS` → the feature description from step 1
+   - `$PLAN_PATH` → the path to the plan file created during the plan phase
+
+3. **Evaluate the gate** — If the phase has a `gate:`, verify the gate conditions are met before proceeding to the next phase. If the gate fails, retry or ask the user for guidance (depending on orchestrator personality).
+
+4. **Track state** — Remember the plan file path when created, track which phases have completed, note key decisions.
+
+5. **Handle signals** — If the phase has a `signal:` instead of a `skill:`, output that signal (e.g., `<promise>DONE</promise>`).
+
+### Variable threading
+
+- After the plan phase completes, scan `docs/plans/` for the most recently created plan file and store its path as `$PLAN_PATH`.
+- Pass `$PLAN_PATH` to subsequent phases that reference it in their `args:`.
+
+## Step 6: Review preferences
+
+When invoking `/ce:review`, pass the orchestrator's `review-preferences` and `synthesis` configuration as context:
+
+- **min-reviewers** — Minimum number of reviewers to spawn
+- **require-categories** — Categories that must be represented (warn if no reviewer available)
+- **prefer-categories** — Categories to include if available
+- **synthesis.lens** — Pass this to the synthesis reviewer to shape how findings are weighted
+
+These preferences guide reviewer selection but don't override the existing ce:review selection logic — they add constraints on top of it.
+
+## Step 7: Model selection
+
+Orchestrators define two model fields:
+
+- **`orchestrator-model`** — The model for the orchestrator itself (the main conversation thread). `inherit` means use the session model.
+- **`agent-model`** — The default model for skills and subagents the orchestrator spawns.
+
+Per-phase `model:` overrides take precedence over `agent-model`. Resolution order:
+
+1. Phase-level `model:` (if specified)
+2. Orchestrator-level `agent-model:` (if specified)
+3. Session model (inherit)
+
+When spawning Agent subagents, pass the resolved model. When invoking skills in the main conversation (e.g., `/ce:plan`), the orchestrator-model applies since skills run in the main thread.
+
+## Step 8: Completion
+
+When all phases are done, summarize the workflow:
+- Which phases ran (and which were skipped, with reasons)
+- Key decisions made along the way
+- Any learnings captured in the compound phase
+
+Communicate completion in the orchestrator's voice.
+
+## Available Orchestrators
+
+To see what's installed, run:
+
+```bash
+ls $PLUGIN_DIR/orchestrators/*.md 2>/dev/null | xargs -I{} basename {} .md
+```
+
+If no orchestrators are found, run `/ce:refresh` to sync from configured sources.

plugins/compound-engineering/skills/ce-run/references/orchestrator-registry.yaml

@@ -0,0 +1,19 @@
+# Orchestrator Registry
+#
+# Orchestrator .md files live in external repos and are synced into
+# orchestrators/ via /ce:refresh. Each orchestrator's phases, review
+# preferences, and personality are defined in its own frontmatter
+# and body.
+#
+# To add a custom orchestrator:
+#   1. Add the .md file to your repo (with type: orchestrator in frontmatter)
+#   2. Run /ce:refresh to pull it in
+#   3. Use /ce:run <name> to invoke it
+
+# === External orchestrator sources ===
+
+sources:
+  - name: ce-default
+    repo: JumpstartLab/ce-reviewers-jsl
+    branch: main
+    path: orchestrators

plugins/compound-engineering/skills/lfg/SKILL.md

@@ -1,32 +1,7 @@
 ---
 name: lfg
-description: Full autonomous engineering workflow
+description: "Full autonomous engineering workflow. Shortcut for /ce:run lfg."
 argument-hint: "[feature description]"
-disable-model-invocation: true
 ---
 
-CRITICAL: You MUST execute every step below IN ORDER. Do NOT skip any required step. Do NOT jump ahead to coding or implementation. The plan phase (step 2) MUST be completed and verified BEFORE any work begins. Violating this order produces bad output.
-
-1. **Optional:** If the `ralph-loop` skill is available, run `/ralph-loop:ralph-loop "finish all slash commands" --completion-promise "DONE"`. If not available or it fails, skip and continue to step 2 immediately.
-
-2. `/ce:plan $ARGUMENTS`
-
-   GATE: STOP. Verify that the `ce:plan` workflow produced a plan file in `docs/plans/`. If no plan file was created, run `/ce:plan $ARGUMENTS` again. Do NOT proceed to step 3 until a written plan exists. **Record the plan file path** — it will be passed to ce:review in step 4.
-
-3. `/ce:work`
-
-   GATE: STOP. Verify that implementation work was performed - files were created or modified beyond the plan. Do NOT proceed to step 4 if no code changes were made.
-
-4. `/ce:review mode:autofix plan:<plan-path-from-step-2>`
-
-   Pass the plan file path from step 2 so ce:review can verify requirements completeness.
-
-5. `/compound-engineering:todo-resolve`
-
-6. `/compound-engineering:test-browser`
-
-7. `/compound-engineering:feature-video`
-
-8. Output `<promise>DONE</promise>` when video is in PR
-
-Start with step 2 now (or step 1 if ralph-loop is available). Remember: plan FIRST, then work. Never skip the plan.
+Run `/ce:run lfg $ARGUMENTS`

plugins/compound-engineering/skills/refresh/SKILL.md

@@ -1,11 +1,11 @@
 ---
 name: ce:refresh
-description: "Sync reviewer personas from external Git repos into the local plugin. Reads sources from reviewer-registry.yaml, fetches .md files, and places them in agents/review/. Use when setting up the plugin for the first time, after updating your reviewer repo, or to pull in new reviewer personas."
+description: "Sync reviewer personas and orchestrator definitions from external Git repos into the local plugin. Use when setting up the plugin for the first time, after updating your repos, or to pull in new reviewers or orchestrators."
 ---
 
-# Refresh Reviewers
+# Refresh Reviewers & Orchestrators
 
-Syncs reviewer persona files from external Git repositories into the plugin's `agents/review/` directory.
+Syncs reviewer persona files and orchestrator definitions from external Git repositories into the plugin.
 
 ## Step 1: Locate plugin
 
@@ -21,56 +21,87 @@ Fall back to relative path if not found (e.g., running from source repo):
 PLUGIN_DIR="${PLUGIN_DIR:-plugins/compound-engineering}"
 ```
 
+Ensure the orchestrators directory exists:
+
+```bash
+mkdir -p "$PLUGIN_DIR/orchestrators"
+```
+
 ## Step 2: Interactive source configuration
 
-Read the user's source config at `~/.config/compound-engineering/reviewer-sources.yaml`. If it doesn't exist, the sync script will create it on first run — skip this step and go directly to Step 3.
+Read the user's reviewer source config at `~/.config/compound-engineering/reviewer-sources.yaml`. If it doesn't exist, the sync script will create it on first run — skip this step and go directly to Step 3.
 
 If the file exists, parse it and present the current sources to the user like this:
 
 List the current sources, then present three options using AskUserQuestion:
 
 ```
-Current sources:
-  - jsl-reviewers (JumpstartLab/ce-reviewers-jsl@main)
-  - ce-default (JumpstartLab/ce-reviewers@main)
+Current reviewer sources:
+  - jsl-reviewers (JumpstartLab/ce-reviewers-jsl@main, path: reviewers)
+  - ce-default (JumpstartLab/ce-reviewers@main, path: reviewers)
+
+Current orchestrator sources:
+  - jsl-orchestrators (JumpstartLab/ce-reviewers-jsl@main, path: orchestrators)
 
 1. Sync now
-2. Edit config file
+2. Edit config files
 3. Or type a request (e.g., "add owner/repo", "remove ce-default")
 ```
 
 Handle the response:
 - **1 / Sync now** — proceed to Step 3.
-- **2 / Edit config file** — open `~/.config/compound-engineering/reviewer-sources.yaml` via Bash: `${EDITOR:-$(command -v code 2>/dev/null || echo nano)} ~/.config/compound-engineering/reviewer-sources.yaml`. After the editor, re-read the file and present the menu again.
-- **Anything else** — interpret as a natural language request to modify the config (add a source, remove one, change a branch, add an except entry, etc.). Edit the YAML accordingly, then present the menu again.
+- **2 / Edit config files** — open both config files in editor. After editing, re-read and present the menu again.
+- **Anything else** — interpret as a natural language request to modify one or both configs. Edit accordingly, then present the menu again.
 
-## Step 3: Sync
+## Step 3: Sync reviewers
 
-Run the sync script:
+Run the sync script for reviewers:
 
 ```bash
 bash "$PLUGIN_DIR/skills/refresh/sync-reviewers.sh" \
   "$PLUGIN_DIR/skills/ce-review/references/reviewer-registry.yaml" \
   "$PLUGIN_DIR/agents/review"
 ```
 
-## Step 4: Show results
+## Step 4: Sync orchestrators
 
-The script writes a summary to `~/.config/compound-engineering/last-refresh-summary.md`. Read that file and **output its contents to the user exactly as written**. The summary is already formatted as markdown — do not summarize, paraphrase, or reformat it. Just show it.
+Run the same sync script for orchestrators, using the orchestrator registry and output directory:
+
+```bash
+bash "$PLUGIN_DIR/skills/refresh/sync-reviewers.sh" \
+  "$PLUGIN_DIR/skills/ce-run/references/orchestrator-registry.yaml" \
+  "$PLUGIN_DIR/orchestrators" \
+  orchestrator
+```
+
+**Note:** The sync script's third argument (`orchestrator`) tells it to use `orchestrator-sources.yaml` as the user config instead of `reviewer-sources.yaml`. It fetches `.md` files from configured sources regardless of content type.
+
+## Step 5: Show results
+
+The sync script writes summaries to `~/.config/compound-engineering/`:
+- `last-reviewer-refresh-summary.md` — reviewer sync results
+- `last-orchestrator-refresh-summary.md` — orchestrator sync results
+
+Read both files and **output their contents to the user exactly as written**. The summaries are already formatted as markdown — do not summarize, paraphrase, or reformat them.
+
+If only one summary file exists (e.g., no orchestrator sources configured yet), show just the one that exists.
 
 ## Source YAML Format
 
+Both reviewer and orchestrator source configs use the same format:
+
 ```yaml
 sources:
-  - name: my-reviewers
+  - name: my-source
     repo: username/repo-name
     branch: main
-    path: .
+    path: reviewers
 
-  - name: community-reviewers
+  - name: another-source
     repo: SomeOrg/ce-reviewers
+    path: orchestrators
     except:
-      - kieran-python-reviewer
+      - name-to-skip
 ```
 
 | Field | Required | Default | Description |
@@ -79,7 +110,7 @@ sources:
 | `repo` | yes | — | GitHub owner/repo |
 | `branch` | no | `main` | Branch to fetch from |
 | `path` | no | `.` | Directory in the repo containing .md files |
-| `except` | no | `[]` | Reviewer filenames (without .md) to skip |
+| `except` | no | `[]` | Filenames (without .md) to skip |
 
 ## Conflict Resolution
 

plugins/compound-engineering/skills/refresh/sync-reviewers.sh

@@ -1,43 +1,46 @@
 #!/usr/bin/env bash
 #
-# sync-reviewers.sh — Fetch reviewer .md files from configured external repos
+# sync-reviewers.sh — Fetch .md files from configured external repos
 #
-# Usage: ./sync-reviewers.sh <default-registry-yaml> <output-dir>
+# Usage: ./sync-reviewers.sh <default-registry-yaml> <output-dir> [<config-name>]
 #
-# Checks for a user-level config at ~/.config/compound-engineering/reviewer-sources.yaml.
+# config-name defaults to "reviewer" — set to "orchestrator" for orchestrator sync.
+# Checks for a user-level config at ~/.config/compound-engineering/<config-name>-sources.yaml.
 # If it doesn't exist, creates it from the default registry. Then reads sources from
 # the user config, fetches .md files, and writes them to the output directory.
 # First-listed source wins on filename conflicts (processed in reverse order).
 
 set -u
 
-DEFAULT_REGISTRY="${1:?Usage: sync-reviewers.sh <default-registry-yaml> <output-dir>}"
-OUTPUT_DIR="${2:?Usage: sync-reviewers.sh <default-registry-yaml> <output-dir>}"
+DEFAULT_REGISTRY="${1:?Usage: sync-reviewers.sh <default-registry-yaml> <output-dir> [<config-name>]}"
+OUTPUT_DIR="${2:?Usage: sync-reviewers.sh <default-registry-yaml> <output-dir> [<config-name>]}"
+CONFIG_NAME="${3:-reviewer}"
 
 USER_CONFIG_DIR="$HOME/.config/compound-engineering"
-USER_CONFIG="$USER_CONFIG_DIR/reviewer-sources.yaml"
+USER_CONFIG="$USER_CONFIG_DIR/${CONFIG_NAME}-sources.yaml"
 
 # --- First-run: seed user config from default ---
 if [ ! -f "$USER_CONFIG" ]; then
   mkdir -p "$USER_CONFIG_DIR"
-  cat > "$USER_CONFIG" << 'YAML'
-# Reviewer Sources
+  CONFIG_LABEL="$(echo "$CONFIG_NAME" | sed 's/.*/\u&/')"
+  cat > "$USER_CONFIG" << YAML
+# ${CONFIG_LABEL} Sources
 #
-# Configure which repos to pull reviewer personas from.
-# Run /ce:refresh to sync reviewers after editing this file.
+# Configure which repos to pull ${CONFIG_NAME} definitions from.
+# Run /ce:refresh to sync after editing this file.
 #
 # Sources listed first have higher priority. If two sources have
 # a file with the same name, the first source's version is kept.
 #
 # To add a source, copy this template and uncomment it:
 #
-#  - name: my-reviewers
+#  - name: my-${CONFIG_NAME}s
 #    repo: owner/repo-name
 #    branch: main
-#    path: reviewers
+#    path: ${CONFIG_NAME}s
 #    except:
-#      - reviewer-to-skip
-#      - another-reviewer-to-skip
+#      - name-to-skip
+#      - another-to-skip
 
 YAML
   # Append the sources from the default registry
@@ -54,7 +57,7 @@ for line in lines:
         print(line, end='')
 " >> "$USER_CONFIG"
   echo "Created $USER_CONFIG"
-  echo "Edit this file to add your own reviewer repos, then run /ce:refresh again."
+  echo "Edit this file to add your own ${CONFIG_NAME} repos, then run /ce:refresh again."
   echo ""
 fi
 
@@ -185,8 +188,8 @@ sources_json=$(parse_sources)
 num_sources=$(echo "$sources_json" | python3 -c "import json,sys; print(len(json.load(sys.stdin)))")
 
 if [ "$num_sources" -eq 0 ]; then
-  echo "No external reviewer sources configured."
-  echo "Add sources to reviewer-registry.yaml and run again."
+  echo "No external ${CONFIG_NAME} sources configured."
+  echo "Add sources to ${CONFIG_NAME}-sources.yaml and run again."
   exit 0
 fi
 
@@ -301,10 +304,11 @@ done
 total=$((added + updated + unchanged))
 
 # Write summary to a file that Claude can Read and show verbatim
-SUMMARY_FILE="${USER_CONFIG_DIR}/last-refresh-summary.md"
+CONFIG_LABEL="$(echo "$CONFIG_NAME" | sed 's/.*/\u&/')"
+SUMMARY_FILE="${USER_CONFIG_DIR}/last-${CONFIG_NAME}-refresh-summary.md"
 
 {
-echo "# Reviewer Refresh Summary"
+echo "# ${CONFIG_LABEL} Refresh Summary"
 echo ""
 
 # Per-source report — compact comma-separated format
@@ -339,7 +343,7 @@ for (( i=0; i<num_sources; i++ )); do
   echo ""
 done
 
-echo "${total} reviewers synced. ${added} added, ${updated} updated, ${unchanged} unchanged."
+echo "${total} ${CONFIG_NAME}s synced. ${added} added, ${updated} updated, ${unchanged} unchanged."
 [ "$skipped" -gt 0 ] && echo "${skipped} excluded by config."
 [ "$conflicts" -gt 0 ] && echo "${conflicts} conflicts resolved (first source wins)."
 } > "$SUMMARY_FILE"