agentleFS
TidalPaladinTidalPaladin/skills/notify-wake/SKILL.md

notify-wake

Skill1 starsChanged 24 days agoApache-2.0View on GitHub

Set up or inspect durable task wakeups. Use automatically only for operations estimated before launch to take strictly more than 10 minutes, or on explicit request.

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 8 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep source completion independent of Codex availability.

    line 9
  2. Distinctive:

    Read that reference for delivery, owned goal waits, recovery, or cutover diagnosis.

    line 23
  3. Distinctive:

    Keep transport, authority capture, goal lifecycle, delivery, and reconciliation in the shared runtime.

    line 31
  4. Very specific:

    Use --format json for automation.

    line 70
  5. Distinctive:

    Before launch, define the exact target, attention predicate, deadline, and fallback.

    line 75
  6. Very specific:

    Keep research_compatibility as the default.

    line 80
  7. Very specific:

    Never set model or effort on a root turn/start.

    line 84
  8. Distinctive:

    Use Luna 5.6 medium only for read-only checks or dedicated relays.

    line 86
  9. Very specific:

    Every wake must report Elapsed before notification: <seconds> seconds.

    line 88
  10. Distinctive:

    Keep secrets, logs, remote content, and artifacts out of wake input.

    line 92
  11. Distinctive:

    If no reliable event source or exact watcher exists, report the missing capability.

    line 94
  12. Distinctive:

    Do not promise automatic continuation or substitute model polling without authorization.

    line 95

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Notify and Wake

Wake the originating task when an exact operation requires attention. Keep source completion independent of Codex availability.

Invocation and Boundaries

Automatic use requires a defensible prelaunch estimate of strictly more than 10 minutes. Exactly 10 minutes and unknown runtimes use ordinary waits or bounded status checks. Explicit requests bypass the duration gate. Design and debugging need no runtime threshold.

Invocation does not authorize the underlying operation, credentials, services, or wider permissions. Honor existing task authority without asking again.

For a local Unix command or Linux PID, use the locked CLI below. For another provider or adapter changes, read design patterns. Read that reference for delivery, owned goal waits, recovery, or cutover diagnosis.

Source and Runtime Contract

The canonical source is notify-wake/ in TidalPaladin/skills. Resolve the exact containing commit with git rev-parse HEAD. Repository adapters must pin package version 1.0.0 from that full SHA. They may own events, trusted prompts, registered roots, controllers, and retry timing. Keep transport, authority capture, goal lifecycle, delivery, and reconciliation in the shared runtime.

The runtime supports Python 3.11 through 3.14 and requires Codex app-server 0.146.0 or a later schema-conforming release. It exposes:

  • WakeContext, WakeRequest, DeliveryOutcome, and NotifyWaitLease.
  • app-server socket discovery, transport, authority capture, delivery, and reconciliation.
  • DeliveryPolicy.RESEARCH_COMPATIBILITY, which is the default.
  • explicit DeliveryPolicy.STRICT.
  • enter_notify_wait() and deliver_wake().

research_compatibility permits the non-atomic transitions documented in design patterns. It does not permit old state or response formats.

State contract version 2 is mandatory. Global watches live under ${CODEX_HOME:-$HOME/.codex}/notify-wake/v2/. Research queues live under the registered root's .notify-wake/v2/. Do not parse, migrate, requeue, or conditionally support version-1 contexts, notifications, ledgers, or response shapes. Preserve old files only as inert audit evidence. Report version mismatches as unsupported notify-wake contract; cutover required.

Bundled Local-Process CLI

The synced skill contains a locked uv project. Run its preflight before a local command or existing Linux PID:

notify_wake_skill="${CODEX_HOME:-$HOME/.codex}/skills/notify-wake"
uv run --project "$notify_wake_skill" --locked \
  python "$notify_wake_skill/scripts/notify_wake.py" preflight --format json

Public commands:

  • preflight checks Codex 0.146.0, the managed daemon, authority, and the default delivery policy.
  • run --timeout-seconds N [--wake-on always|failure] [--evidence ABS] -- COMMAND... registers an owned process before releasing it.
  • attach --pid PID --timeout-seconds N [--expect-start-ticks TICKS] [--evidence ABS] captures a Linux pidfd and /proc start time. It never signals the attached process.
  • wait WATCH_ID enters an owned persistent-goal wait after the watch is durably armed.
  • status WATCH_ID reads one exact v2 watch.
  • reconcile WATCH_ID reconciles one exact uncertain request boundary.

For a watched uv command in another project, prefix it with env -u VIRTUAL_ENV. This prevents notifier-environment leakage.

Use --format json for automation. The CLI returns 0 for success, 1 when durable state requires attention, and 2 for a runtime or state error.

Completion and Recovery

Before launch, define the exact target, attention predicate, deadline, and fallback. Let the bundled controller capture authority, persist the watch, bind the target, and reconcile registration races. Return control only after the controller is durable. Enter an owned goal wait when no immediate work remains and the goal API permits it.

Keep research_compatibility as the default. Preserve authority-mismatch blockers, manual goal blocks, and uncertain delivery state. Never retry an uncertain request until authoritative history proves it absent.

Never set model or effort on a root turn/start. The root model owns launches, recovery, goal changes, and substantive decisions. Use Luna 5.6 medium only for read-only checks or dedicated relays.

Every wake must report Elapsed before notification: <seconds> seconds. Measure from operation start to the terminal event, or from observation start when the earlier start is unavailable. Keep this value fixed across retries. Send validated identifiers, status, times, and evidence paths only. Keep secrets, logs, remote content, and artifacts out of wake input.

If no reliable event source or exact watcher exists, report the missing capability. Do not promise automatic continuation or substitute model polling without authorization.

Report the runtime estimate, exact target, armed adapter, watch location, attention predicate, deadline, and fallback.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.