agentleFS
D-SheyD-Shey/autoharness/.agents/skills/autonomous-backlog-driver/SKILL.md

autonomous-backlog-driver

Skill10 starsChanged todayMITView on GitHub

Continuous task execution loop over followups with regression rollback guards and safety budgets

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    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:

    Use when running unattended or semi-autonomous development cycles that iteratively work down technical debt, open followups, and bug …

    line 9
  2. Distinctive:

    Test regression sentinel: run baseline verification before task execution.

    line 17
  3. Distinctive:

    Create or assign task plan.

    line 19
  4. Distinctive:

    If tests fail, invoke rollback to baseline tree state and record a ruling.

    line 22
  5. Distinctive:

    If tests pass and critic accepts, mark followup as done, record lesson, and commit changes.

    line 23
  6. Very specific:

    Run driver in --dry-run mode: verify candidate selection order without touching files.

    line 27
  7. Very specific:

    Run driver with --max-tasks 1: verify single task processing, state update, and graceful termination.

    line 28
  8. Distinctive:

    Verify regression guard: simulate a failing test, confirm immediate rollback of changes.

    line 29

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

autonomous-backlog-driver

When to use

Use when running unattended or semi-autonomous development cycles that iteratively work down technical debt, open followups, and bug backlogs without manual task-by-task prompting.

Steps

  1. Load pending followups from .autoharness/followups.json where status == "open".
  2. Prioritize items based on explicit order or creation timestamp.
  3. Establish a safety budget:
    • Maximum tasks per drive session (default: 3).
    • Maximum execution time per task (timeout guard).
    • Test regression sentinel: run baseline verification before task execution.
  4. For each selected item:
    • Create or assign task plan.
    • Execute task implementation.
    • Run verification checks.
    • If tests fail, invoke rollback to baseline tree state and record a ruling.
    • If tests pass and critic accepts, mark followup as done, record lesson, and commit changes.
  5. Persist driver session state in .autoharness/drive_state.json.

How to verify

  1. Run driver in --dry-run mode: verify candidate selection order without touching files.
  2. Run driver with --max-tasks 1: verify single task processing, state update, and graceful termination.
  3. Verify regression guard: simulate a failing test, confirm immediate rollback of changes.

Facts found here

  • Outer loop automation requires atomic state persistence and hard budget limits to prevent runaway resource consumption.

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.