summaryrefslogtreecommitdiff
path: root/Omni/Agent
diff options
context:
space:
mode:
Diffstat (limited to 'Omni/Agent')
-rw-r--r--Omni/Agent/DESIGN.md104
-rwxr-xr-xOmni/Agent/monitor-worker.sh47
-rwxr-xr-xOmni/Agent/start-worker.sh2
3 files changed, 152 insertions, 1 deletions
diff --git a/Omni/Agent/DESIGN.md b/Omni/Agent/DESIGN.md
new file mode 100644
index 0000000..c3fa792
--- /dev/null
+++ b/Omni/Agent/DESIGN.md
@@ -0,0 +1,104 @@
+# Multi-Agent System 2.0 Design
+
+**Goal:** Replace the current bash-script based worker system (`start-worker.sh`, etc.) with a robust, type-safe Haskell application `Omni/Agent.hs`.
+
+## 1. CLI Interface
+
+The `agent` command (compiled from `Omni/Agent.hs`) will provide a unified interface for managing workers.
+
+```bash
+agent start <name> [--background] # Start a worker (foreground by default, background with flag)
+agent stop <name> # Stop a background worker
+agent status # List all workers and their status
+agent log <name> [-f] # View/tail worker logs
+agent harvest # Harvest task updates from all workers
+agent sync # Sync local state with live (helper)
+```
+
+## 2. Module Structure (`Omni/Agent/`)
+
+We will refactor the bash logic into Haskell modules:
+
+- **Omni.Agent** (`Omni/Agent.hs`): Main entry point and CLI parsing (Docopt).
+- **Omni.Agent.Core**: Core data types and state management.
+- **Omni.Agent.Worker**: The worker loop logic (sync, claim, work, submit).
+- **Omni.Agent.Git**: Git operations (worktree, branch, merge, commit).
+- **Omni.Agent.Process**: Process management (PID files, signals).
+- **Omni.Agent.Log**: Log streaming and filtering (the "monitor" logic).
+
+## 3. Data Types
+
+```haskell
+data WorkerStatus
+ = Idle
+ | Syncing
+ | Working TaskId
+ | Submitting TaskId
+ | Error Text
+ deriving (Show, Eq, Generic)
+
+data Worker = Worker
+ { workerName :: Text
+ , workerPid :: Maybe Int
+ , workerStatus :: WorkerStatus
+ , workerPath :: FilePath
+ }
+```
+
+## 4. Implementation Details
+
+### 4.1 Worker Loop (`agent start`)
+The Haskell implementation should replicate the logic of `start-worker.sh` but with better error handling and logging.
+
+1. **Setup**: Ensure worktree exists (or create it).
+2. **Loop**:
+ - `Git.syncWithLive`
+ - `Task.sync`
+ - `task <- Task.findReady`
+ - If `task`:
+ - `Task.claim task`
+ - `Git.checkoutTaskBranch task`
+ - `Amp.execute prompt`
+ - `Git.commit`
+ - `Git.checkoutBase`
+ - `Task.submitReview task`
+ - Else: `sleep 60`
+
+### 4.2 Process Management
+- Store PIDs in `.tasks/workers/<name>.pid`.
+- `agent stop` sends SIGTERM to the PID.
+- `agent status` checks if PID is alive.
+
+### 4.3 Logging
+- Continue writing raw Amp logs to `_/llm/amp.log` in the worker directory.
+- `agent log` reads this file and applies the filtering logic (currently in `monitor-worker.sh` jq script) using Haskell (Aeson).
+- **Requirement:** Output must include timestamps for every event. Extract the `timestamp` field from the JSON log and format it (e.g., `[HH:MM:ss] 🤖 THOUGHT: ...`).
+
+### 4.4 Harvesting
+- Iterate over `.tasks/workers/` or `git worktree list`.
+- For each worker, extract `.tasks/tasks.jsonl` via `git show`.
+- Run `Task.import`.
+
+## 5. Migration Strategy
+
+1. **Parallel Existence**: Keep bash scripts while developing Haskell version.
+2. **Feature Parity**: Ensure `agent start` works exactly like `start-worker.sh`.
+3. **Cutover**: Update `WORKER_AGENT_GUIDE.md` to use `agent` command.
+4. **Cleanup**: Delete bash scripts.
+
+## 6. Testing Plan
+
+### 6.1 Unit Tests (`Omni/Agent/Test.hs`)
+- Test `Git` module commands (mocked).
+- Test `Log` filtering logic.
+- Test CLI argument parsing.
+
+### 6.2 Integration Tests
+- Create a temporary test repo.
+- Spawn a worker.
+- Mock `amp` binary (simple script that `echo "done"`).
+- Verify task moves from Open -> InProgress -> Review.
+
+## 7. References
+- `Omni/Agent/start-worker.sh` (Current implementation)
+- `Omni/Task.hs` (Task manager integration)
diff --git a/Omni/Agent/monitor-worker.sh b/Omni/Agent/monitor-worker.sh
new file mode 100755
index 0000000..2638e2d
--- /dev/null
+++ b/Omni/Agent/monitor-worker.sh
@@ -0,0 +1,47 @@
+#!/usr/bin/env bash
+set -e
+
+# Omni/Agent/monitor-worker.sh
+# Monitors the worker agent's activity by filtering the debug log.
+# Usage: ./Omni/Agent/monitor-worker.sh [worker-directory-name]
+
+WORKER_NAME="${1:-omni-worker-1}"
+REPO_ROOT="$(git rev-parse --show-toplevel)"
+WORKER_PATH="$REPO_ROOT/../$WORKER_NAME"
+LOG_FILE="$WORKER_PATH/_/llm/amp.log"
+
+if [ ! -f "$LOG_FILE" ]; then
+ echo "Waiting for log file at $LOG_FILE..."
+ while [ ! -f "$LOG_FILE" ]; do sleep 1; done
+fi
+
+echo "Monitoring Worker Agent in '$WORKER_PATH'..."
+echo "Press Ctrl+C to stop."
+echo "------------------------------------------------"
+
+# Tail the log and use jq to parse/filter relevant events
+# We handle JSON parse errors gracefully (in case of partial writes)
+tail -f "$LOG_FILE" | grep --line-buffered "^{" | jq -R -r '
+try (
+ fromjson |
+ if .message == "executing 1 tools in 1 batch(es)" then
+ "🤖 THOUGHT: Planning tool execution (" + (.batches[0][0] // "unknown") + ")"
+ elif .message == "Tool Bash - checking permissions" then
+ empty
+ elif .message == "Tool Bash permitted - action: allow" then
+ "🔧 TOOL: Bash command executed"
+ elif .toolName != null and .message == "Processing tool completion for ledger" then
+ "✅ TOOL: " + .toolName + " completed"
+ elif .message == "ide-fs" and .method == "readFile" then
+ "📂 READ: " + .path
+ elif .message == "System prompt build complete (no changes)" then
+ "🧠 THINKING..."
+ elif .message == "System prompt build complete (first build)" then
+ "🚀 STARTING new task context"
+ elif .level == "error" then
+ "❌ ERROR: " + .message
+ else
+ empty
+ end
+) catch empty
+'
diff --git a/Omni/Agent/start-worker.sh b/Omni/Agent/start-worker.sh
index d005156..25f325c 100755
--- a/Omni/Agent/start-worker.sh
+++ b/Omni/Agent/start-worker.sh
@@ -117,7 +117,7 @@ Context:
"
mkdir -p _/llm
- "$AMP_BIN" --log-file "_/llm/amp.log" --dangerously-allow-all -x "$PROMPT"
+ "$AMP_BIN" --log-level debug --log-file "_/llm/amp.log" --dangerously-allow-all -x "$PROMPT"
AGENT_EXIT_CODE=$?