@hasna/loops 0.3.47 → 0.3.49

package/README.md

@@ -5,7 +5,7 @@ OpenLoops is a local CLI and daemon for persistent loops and workflows: schedule
 It supports deterministic command loops, JSON-defined workflows, and guarded CLI adapters for headless coding agents:
 
 - `claude`
-- `cursor agent` or `agent`
+- `agent` (Cursor Agent CLI)
 - `codewith exec`
 - `aicopilot run`
 - `opencode run`
@@ -315,6 +315,38 @@ loops routes show <work-item-id>
 loops routes invocations
 ```
 
+For Hasna OSS task-created routing, use a deterministic drain instead of tmux
+dispatch:
+
+```bash
+loops routes schedule todos-task oss-task-route-drain \
+  --every 5m \
+  --todos-project "$HOME/.hasna/loops" \
+  --project-path-prefix /home/hasna/workspace/hasna/opensource \
+  --tags auto:route \
+  --provider codewith \
+  --auth-profile-pool account004,account005,account006 \
+  --add-dir "$HOME/.hasna/todos,$HOME/.hasna/loops" \
+  --project-group oss \
+  --max-dispatch 2 \
+  --scan-limit 5000 \
+  --max-active-per-project 1 \
+  --max-active-per-project-group 4 \
+  --max-active 12 \
+  --worktree-mode required \
+  --evidence-dir "$HOME/.hasna/loops/reports/oss-task-route-drain" \
+  --compact
+```
+
+Only route tasks that explicitly opt in with `auto:route`, `route_enabled=true`,
+or `automation.allowed=true`. Use Codewith account pools, required worktrees,
+max-active throttles, and bounded evidence directories. Do not dispatch or paste
+task prompts into tmux panes. Keep `--scan-limit` large enough for the local
+ready-task backlog; otherwise low-priority or newly-created tasks can sit beyond
+the scanned window. Generated one-shot task/event route workflow specs are
+archived automatically when their workflow run reaches a terminal state;
+workflow run history and manifests are preserved.
+
 Workflow run manifests are written under
 `.hasna/loops/runs/<project-slug>/<subject-key>/<run-id>/manifest.json`.
 `subject-key` is a safe derived path segment, not the raw subject reference.
@@ -499,8 +531,8 @@ The adapters intentionally use provider command surfaces instead of pretending e
 - Claude uses `claude -p --output-format json` and safe-mode/local setting sources by default.
 - Codewith uses `codewith --ask-for-approval never exec --json --ephemeral --skip-git-repo-check`, with `--add-dir` for explicit extra writable directories.
 - AI Copilot and OpenCode use `run --format json --pure`.
-- Cursor is CLI-first for now via standalone `agent -p` when available, with `cursor agent -p` as a compatibility fallback; treat output as less stable until a stronger public SDK contract is selected.
-- Codex uses `codex exec --json --ephemeral --skip-git-repo-check`, with `--add-dir` for explicit extra writable directories where supported.
+- Cursor is CLI-first for now via the standalone `agent -p` binary. OpenLoops no longer falls back to `cursor agent`; install the standalone Cursor Agent CLI so preflight and scheduled runs use the same executable.
+- Codex uses `codex --ask-for-approval never exec --json --ephemeral --skip-git-repo-check`, with `--add-dir` for explicit extra writable directories where supported.
 - Agent prompts are sent through child stdin instead of argv so prompt bodies do not appear in process listings.
 - When `--account` or a step `account` is set, OpenLoops resolves `accounts env <profile> --tool <tool>` before spawning the target, strips inherited tool home/API-key variables, and applies the selected profile only to that process. Missing account profiles fail before the provider binary receives the prompt.
 - `--auth-profile` and step `authProfile` are provider-native auth selectors. They currently apply to Codewith and are passed to Codewith as `--auth-profile <name>` before `exec`; they do not call OpenAccounts.