AgentStack
SKILL verified MIT Self-run

Flow Auto

skill-edouard-claude-pi-flow-skills-flow-auto · by edouard-claude

Orchestrator that loops through every ready-for-dev story in the sprint in order, chaining /flow-story -> /flow-dev

No reviews yet
0 installs
10 views
0.0% view→install

Install

$ agentstack add skill-edouard-claude-pi-flow-skills-flow-auto

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Flow Auto? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

flow-auto — multi-story orchestration

This skill is not executed by the LLM in the classic sense. It is documentation for the companion run.mjs script that drives Pi in --print --no-session mode to loop over the sprint stories.

Concept

A Pi skill is text instructions, not a program. To orchestrate multiple Pi invocations (including /clear between each), you need an external script that runs Pi in non-interactive mode. Since v0.9.0, that script is a single self-contained ESM bundle that runs on the Node already bundled with Pi — no bash, python, uvx, or jq needed.

The companion script

~/.pi/agent/git/github.com/edouard-claude/pi-flow-skills/skills/flow/flow-auto/run.mjs

For each ready-for-dev story (or backlog with satisfied dependencies):

  1. pi --print --no-session "/flow-story "
  2. pi --print --no-session "/flow-dev "
  3. pi --print --no-session "/flow-review " (up to 3 dev↔review cycles)
  4. pi --print --no-session "/flow-commit "

--no-session = ephemeral session = equivalent to /clear between each call.

The loop stops when:

  • No more ready-for-dev or backlog story with satisfied dependencies
  • A step fails (pi exit != 0)
  • A story moves to blocked
  • More than 3 dev↔review cycles on the same story

Launch

~/.pi/agent/git/github.com/edouard-claude/pi-flow-skills/skills/flow/flow-auto/run.mjs [path/sprint-status.yaml]

The #!/usr/bin/env node shebang resolves Node automatically. Default argument: .agents/implementation/sprint-status.yaml.

Requirements

  • sprint-status.yaml up to date (run /flow-sprint first if needed)
  • Skills flow-story, flow-dev, flow-review, flow-commit installed
  • Pi installed (the Node it ships with is enough)

That's it. No bash, no python, no uvx, no jq. The bundle embeds yaml for parsing and a built-in JSON event stream parser.

Environment knobs

  • PI_MODE=text — text output instead of streamed JSON events
  • PI_RAW=1 — raw JSON passthrough (debug)
  • PI_BIN=/path/to/pi — alternate Pi binary
  • NO_COLOR=1 — disable ANSI colors
  • NO_STICKY_HEADER=1 — disable the sticky top header
  • FLOW_PARALLEL=0 — disable all parallel waves (v0.3-equivalent inline behavior)

Guarantees / non-guarantees

Guarantees:

  • Order respected (topological sort of dependencies)
  • A complete story = 1 full Pi cycle (CREATE → DEV → REVIEW → COMMIT)
  • Fresh context between stories (--no-session)
  • Idempotent resume after crash (each phase reads sprint-status and skips if already advanced)

Does NOT guarantee:

  • That /flow-dev finishes without a halt condition (human may be required)
  • That /flow-review approves on first pass (dev↔review cycles allowed up to 3)
  • That commits are pushed (never without explicit user request)

When to prefer manual

  • First story of a project (calibrates expectations)
  • Story with likely halt conditions (ambiguous decisions)
  • Need for human validation between phases

flow-auto is built for batch runs over straightforward stories, not stories requiring close supervision.

Next

  • Run finished without error → /flow-retro if epic complete
  • Halt mid-way → manual resume with /flow-dev or /flow-review
  • Everything done → /flow-help to decide what's next

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.