Commit Graph

10 Commits

Author SHA1 Message Date
jj.wang 5192e04088 PAD-12: per-terminal transport (direct Tailscale SSH or Warpgate, pre-session fallback, pinned host keys, agent key from Vault), one-call win10 launch and one-call collect, remote call timings, agentd ping 2026-10-22 02:59:37 +00:00
jj.wang 6dc7e72ac8 Webhook receiver: Plane deliveries trigger a tick (PAD-7)
agentd webhook-serve (agentd/webhook.py, stdlib): verifies X-Plane-Signature
(HMAC-SHA256 hex of the body, constant-time; 401 on missing/bad), routes by
data.project to a configured tenant (others 202), coalesces deliveries within
3 s into one tick, runs one more tick when deliveries arrive during a tick,
and retries every 15 s for up to 2 min while the lock is held (tick command
exit 75: flock -E 75 in the cron scripts, cli tick on Busy). Ticks run in a
systemd scope so restarting the receiver never kills one. Log:
<log_dir>/webhook.log; GET /healthz.

Config: webhook section (key names only, PLANE_WEBHOOK_SECRET scrubbed from
child envs like the Plane tokens), notify: {kind: none} as a no-op hook
called on every state move; the unused per-project webhook_secret_key is
gone (Plane webhooks are workspace-level). systemd unit for the LXC and
tools/plane-webhook.py to create/rotate the Plane webhook as the workspace
admin (secret straight into the Vault, never printed).
2026-10-21 13:24:04 +00:00
jj.wang 72681db46b Heartbeat: record why each Todo item was not dispatched
Each tick lists skipped items with terminal and reason (concurrency n/m with
the holding runs, serialize group holder, busy_process not started by agentd,
relay down, no route for assignee, start failed) in heartbeat.json "skipped".
A reason is logged (agentd.log / cron.log) only when it is new for that item;
the last set is kept in the state file as "skips".
2026-10-21 13:21:16 +00:00
jj.wang 4f45d368ad Double-start guard: launching marker, probe-and-adopt, fresh-dir relaunch; skip unreachable terminals per tick
Before the launch call the planned run (run id, terminal, handle, bookkeeping) is saved as a launching
marker. If the launch call fails or agentd dies before recording the run, the next tick probes the run
directory (exit.txt / launch.pid / pid.txt): evidence -> adopt the run; nothing -> relaunch with the same
run id in <run_id>-r2 (-r3 ...); probe unreachable -> keep the marker, which holds the terminal slot.
Transports refuse to clean a run directory holding a pid file. A terminal whose relay fails is skipped
for the rest of the tick (heartbeat.down) while other terminals carry on.
2026-10-21 11:54:40 +00:00
jj.wang dc8271a06a warpgate-ssh: remote Linux/macOS terminals through the relay; oregon launched remotely
The dispatcher moves to an LXC without agent CLIs (design 9), so every terminal is
reached through Warpgate. The relay call (agent-access.py run --target <t>) lives in
one helper; the target comes from the terminal config.

os linux/darwin (PosixWarpgateTransport): each primitive is one bash -c call. Files
go up as base64 (chunked above 60k), start_run stages prompt.md + run.sh and launches
in a single round trip; the runner is detached with setsid nohup (pid == pgid == sid,
$! kept in launch.pid); kill signals the process group and session (TERM, then KILL),
skipped once exit.txt exists. macOS detaches with perl POSIX::setsid.

os windows keeps the win10-ps.sh / win10-put.sh helper scripts (target win10) and
schtasks, now through the shared call() helper.

oregon is now warpgate-ssh (target oregon, run_root ~/.gwent-board/agentd-runs);
the local transport stays for co-located setups and tests.
2026-10-21 08:08:01 +00:00
jj.wang f2f7e4308a Per-model service tier default (codex: gpt-6.1-sol -> fast), tier:<x> / tier:none labels
executors.<x>.tier_defaults maps a model to the tier used when no label sets one;
a tier:<name> label sets it explicitly and tier:none drops the tier argument.
2026-10-21 08:08:01 +00:00
jj.wang 1ae44eb705 Allow concurrency 0 to pause a terminal
A paused terminal gets no new runs and is not probed; in-flight runs still finish.
2026-10-21 07:17:29 +00:00
jj.wang efbc5f6e27 Core: cancel moved runs, retry bookkeeping, progress cap, safer classification
- A run whose item leaves In Progress is killed, gets a [worker] cancel comment,
  frees its terminal slot and serialize group and keeps its thread id; if the
  kill fails the slot stays taken and the cancel is retried next tick.
- Transient retries are the same logical round: they do not count toward
  max_rounds_per_thread and do not trigger a max-rounds rebuild.
- New project setting max_progress_rounds (default 12): after that many
  consecutive [progress] rounds the item goes to Waiting (answerer) with a note.
- retry/rebuild markers are consumed only after the run launched, so a failed
  start no longer leaves an In Progress item that is never picked again; a
  rebuild is no longer reported as a retry.
- State is saved after every finished round and every launched run.
- Failure classification: transient patterns only look at the tail, and an
  exit-0 failure is logic (never retried) unless the tail shows a transient error.
- First-line tag: tolerate BOM and a leading code fence / rule line.
- Start comment notes when a handoff summary was attached.
2026-10-21 07:13:18 +00:00
jj.wang 00d42267e5 Harden transports and state files
- Executors and relay helpers get an environment without any configured
  Plane token / webhook secret (design 8: executors hold no Plane credentials).
- Windows runner: Set-Location -LiteralPath, and a failed cd writes exit=97
  instead of running the agent in the wrong directory.
- schtasks launch: escape apostrophes and pass the script path as \"...\" so
  paths with spaces survive PowerShell 5.1 native quoting.
- State and heartbeat files are written temp + fsync + rename.
- Tests for the above plus codex run/resume command shapes and unique run dirs.
2026-10-21 07:10:41 +00:00
jj.wang 31850c4a81 Scheduling core, state store, CLI and unit tests
One tick polls in-flight runs, posts results, classifies the first line and
moves state/assignee; failures get a structured note with transient retry and
backoff, timeouts are killed and handed over, resume failures and long threads
rebuild with a handoff summary. Then the lowest Todo items are dispatched within
terminal concurrency and serialize groups, and a heartbeat is written.
CLI: tick, new, show, comment, state, assign, smoke, check-config.
2026-10-21 07:04:32 +00:00