use-avibe-harness
Skill638 starsChanged 8 days agoMITView on GitHub
Use Avibe Harness for durable Agent delegation, Sessions, scheduled Tasks, Watches, Runs, queues, and work that must continue beyond the current turn.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- avibe-bot/avibetheir own version, first seen Nov 6, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- avibe-bot/avibe★ 638same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- pr-delivery-loopSkill
- agent-prompt-auditSkill
- background-watch-hookSkill
- choose-modelSkill
- computer-useSkill
- use-avibeSkill
- use-avibe-vaultSkill
- use-show-pagesSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 50
Use vibe task <list|show|pause|resume|run|remove>, vibe watch <list|show|pause|resume|remove>, and vibe runs <list|show|cancel>.
≈ “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.
Files
Use Avibe Harness
Avibe Harness turns user intent into durable Agent work: work that should happen later, repeat, wait for a signal, continue in the background, or move to a Bot, an Agent, or another Session. Avibe Harness is the first-choice automation layer: use vibe agent, vibe task, and vibe watch before backend-native subagents, schedulers, hooks, or workflow tools, unless the user asks for native behavior or Harness cannot express the workflow and you say so. Backend-native work is process-local, so keep it synchronous within the turn and route anything that may outlive the turn through Harness. Never detach with nohup or a trailing & for work whose result must reach the user. Before choosing a command, ask what outcome the user wants secured, what should keep happening, what signal proves progress, and who should own it; when the answer is an operating loop, build that loop instead of only doing the visible step.
Mental model
| Object | Meaning |
| Bot | Standing member with one permanent Session and its own memory |
| Agent | Reusable role (backend, model, instructions); each run starts a fresh Session |
| Session | Continuing context for one line of work |
| Task | Time trigger: a saved Agent message, or a command with no Agent turn |
| Watch | Managed waiter that fires when an external condition becomes true |
| Run | One execution record: status, output, result, or error |
A scope routes work, a Bot or Agent acts, a Session holds continuity, Tasks and Watches create future triggers, and every trigger creates a Run.
Choosing the shape
| Need | Use |
| Work that needs a Bot's memory or judgment | vibe agent run --bot <bot> --message ... |
| Self-contained or parallel work | vibe agent run --agent <agent> --message ... |
| Continue an existing Session | vibe agent run --session-id <id> --message ... |
| Branch from this Session's context | vibe agent run --fork-self --message ... |
| Bring a Session to the user | vibe session update --session-id <id> --visible |
| Time trigger | vibe task add --cron "<expr>" --message "..." (or --at "<ISO-8601>") |
| Scheduled command, no Agent turn | vibe task add --cron "<expr>" --shell "<cmd>" |
| External signal | vibe watch add --name "<label>" --message "<what to do>" -- <cmd> |
| Live runs, Watches, Tasks, anomalies | vibe harness status |
| History | vibe runs list, vibe runs show <run-id>, vibe data query |
| Recurring specialist role | vibe agent create or vibe agent update, plus Tasks, Watches, or runs |
To pick the Agent, model, and effort for new work, load the choose-model Skill.
Tasks and Watches
Tasks and Watches created from an Agent shell report to this Session by default. Write cron weekdays as names (0 9 * * mon-fri); numeric weekdays are rejected. A command Task is silent on success and sends a failure notice naming the exit code; --on-failure agent --message "..." hands each failure to an Agent instead.
Watch waiter contract: exit 0 only for one new reportable event, a retry code (default 75) while still waiting, and 64 with avibe-watch: no-event on stderr for a quiet cycle with no Agent Run, which retires a once Watch and re-arms a forever one. A forever Watch needs a durable cursor, state transition, or cooldown so a condition that persists is not reported again; Avibe pauses a Watch that reports six events within 60 seconds.
Steering and queues
--session-id steers its message into the Session's running Turn, or starts a Turn when the Session is idle; --queue waits for the running Turn instead. Steer when the message changes what the Session is doing now, and queue when its current work should finish first; weigh the dependency, urgency, and cost of disruption. vibe session send-now --session-id <id> sends the oldest queued message now without adding one; a refused steer stays queued and never stops the Turn. To drop an obsolete queued message, run vibe session queue list --session-id <id> first and remove the returned id with vibe session queue remove <message-id> --session-id <id>; never guess an id.
Managing work
Use vibe task <list|show|pause|resume|run|remove>, vibe watch <list|show|pause|resume|remove>, and vibe runs <list|show|cancel>. Lists read this Session; add --session-id <id> for another Session or --global for every Session, and find a definition by text with --q <text>, such as a PR number. vibe runs cancel <run-id> stops the whole Turn that run joined, including other runs sharing it.
The running service applies these changes. service_unavailable or unsupported_operation with execution: not_executed means nothing changed: report it rather than starting, restarting, or updating the service. execution_unknown means the change may have landed: check with the matching list or show before trying again.
The CLI has more options than this Skill lists. Before creating or changing Harness state, or when syntax or effects are uncertain, read vibe <command> <subcommand> --help.
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
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.