agentleFS
XORwellXORwell/archeflow/skills/autonomous-mode/SKILL.md

autonomous-mode

Skill0 starsChanged 52 days agoMITView on GitHub

Running ArcheFlow unattended (overnight, batches of tasks): self-organization, progress logging, safe stopping.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Budget exceeded: Stop at limit

    line 49
  2. Distinctive:

    Pause: Stop after current task, resume later

    line 62
  3. Very specific:

    Review: Read .archeflow/session-log.md for progress

    line 64
  4. Very specific:

    Every session writes to .archeflow/session-log.md with per-task entries:

    line 68

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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

Autonomous Mode

ArcheFlow runs can work through several tasks without a person watching each step. The quality gates are LLM reviewers, so two settings stay with the user:

  • Permission mode. Agents inherit the session's permission mode. The skill never asks for a bypass mode; if the user wants fewer prompts, they choose that when starting the session.
  • Merging. A run merges into the base branch only with the user's yes, or when git.auto_merge: true is set in .archeflow/config.yaml and the user confirmed it once in this session (the file can come with the repository; see archeflow:run, Merge). Without either, an unattended run ends with status awaiting_merge and the reviewed work waits on its branch.
  • Tasks from files. Tasks in .archeflow/queue.md can come with the repository. Before the first task, show the user every task line from the file (full text, including done: conditions) and get a yes for the list; a task or line added later needs a new yes. A done: condition that names a command (load_test.sh passes) is run only if the user approved that exact command in this session; otherwise treat the condition as unmet and report it.

Task Queue Formats

Inline:

1. "Fix the login bug" (fast)
2. "Add user profile page" (standard)

From file (.archeflow/queue.md):

- [ ] Fix the login bug | fast
- [ ] Add user profile page | standard | depends: fix login
- [ ] Security audit | thorough | done: Guardian approves AND load_test.sh passes

Tasks with depends: wait for the named task to complete. Tasks with done: have completion criteria checked in the Act phase.

Safety Mechanisms

Automatic Stop Conditions

  • 3 consecutive failures: Something systemic is wrong
  • Test suite broken: archeflow-rollback.sh reverts the run's merge commit; halt
  • Budget exceeded: Stop at limit
  • Shadow escalation: Same shadow detected 3+ times across tasks
  • Destructive action detected: Force push, branch deletion, schema drop

Everything is Reversible

  • Code lives on run branches until merged (with confirmation or git.auto_merge: true)
  • Merges are --no-ff by default (one revertable merge commit per run)
  • Failed tasks leave branches intact for inspection

User Controls

  • Cancel: Kill session, incomplete work stays on branches
  • Pause: Stop after current task, resume later
  • Skip: Move to next task
  • Review: Read .archeflow/session-log.md for progress

Session Log

Every session writes to .archeflow/session-log.md with per-task entries:

  • Workflow, status, cycles, reviewer verdicts
  • Files changed, tests added
  • Branch and commit info
  • Duration and timestamps
  • Session summary at the end

Budget-Aware Scheduling

Budget RemainingAction
> 50%Run at selected workflow level
25-50%Downgrade thorough to standard, standard to fast
< 25%All tasks as fast only
ExhaustedStop, log remaining as skipped

Auto-Resume

On interruption the state is on disk: the event log (.archeflow/events/<run_id>.jsonl), the artifacts, and the run branch (archeflow-git.sh status <run_id>). On the next session, offer to resume with /archeflow:run --start-from <phase> or to start fresh.

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.