agentleFS
fifersheepfifersheep/flow/skills/flow-start/SKILL.md

flow-start

Skill0 starsChanged 3 months agoApache-2.0View on GitHub

Flow phase 1 (setup). Normally invoked via /flow with a new task description. Creates branch/worktree, task folder, workflow.md; recommends full or lite mode.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    MUST be recorded, so later failures are attributable).

    line 36

≈ “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

Flow — setup

Argument: a task description, optionally with a ticket ref (e.g. "BEE-93: …").

  1. Read config. .flow.md at repo root: verify, verify-app, setup (dependency-install command for fresh checkouts), artifacts-dir (default docs/tasks), worktrees, domain-agents, tracker, ## tracker-status (see the plugin's templates/tracker-sync.md) and ## on (per-phase commands — read only by the flow_events.py hook; no skill ever runs them, so never fire one by hand). Missing file → defaults; ask for the verify command once and offer to save a .flow.md.
  2. Check the ticket. Parse the leading identifier out of the argument ("BEE-93: …" → BEE-93); no identifier → skip this step, ticket-id stays blank. With a tracker configured, read the ticket and report its title + current status on one line. A ticket already in a completed or canceled status stops setup — say so and confirm with the user before going further; it usually means the work is already done and this is a re-run.
  3. Recommend a mode. lite = small, well-understood, few files, no design questions; full = everything else. Recommend one with a one-line reason; the user confirms (AskUserQuestion, options full/lite). If the user already stated a mode ("a lite ticket", "run this full"), don't re-ask — announce the mode with your reason and move on.
  4. Create isolation. Branch flow/<slug> from the default branch (worktree at the .flow.md worktrees path if configured; plain branch otherwise). Slug: kebab-case, ≤5 words, from the task description.
  5. Baseline checks. Working tree clean; run setup first if configured (a fresh worktree has no installed dependencies), then verify — record pass/fail in workflow.md Next action (a red baseline is not fatal but MUST be recorded, so later failures are attributable).
  6. Create the task folder <artifacts-dir>/YYYY-MM-DD-<slug>/ and write workflow.md from the plugin's templates/workflow.md, filling every {{placeholder}}. phase: setup, mode as confirmed, ticket free text, ticket-id the bare identifier from step 2 (blank if none). Then move the ticket to the start status per templates/tracker-sync.md and record tracker-synced.
  7. Advance. Set updated, set phase: research (full) or phase: planning (lite), write Next action, announce the transition, then invoke the flow-research or flow-plan skill.

Rules: never start on the default branch; never set force:; if a workflow.md write is blocked by the validator, fix the transition — never work around it. A failed tracker move is recorded and announced, never fatal — full rules in templates/tracker-sync.md.

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.