# Flow Help

> Orientation skill. Detects the current flow phase from .agents/ artifacts and recommends ONE next command among

- **Type:** Skill
- **Install:** `agentstack add skill-edouard-claude-pi-flow-skills-flow-help`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [edouard-claude](https://agentstack.voostack.com/s/edouard-claude)
- **Installs:** 0
- **Category:** [AI & ML](https://agentstack.voostack.com/c/ai-and-ml)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [edouard-claude](https://github.com/edouard-claude)
- **Source:** https://github.com/edouard-claude/pi-flow-skills/tree/main/skills/flow/flow-help

## Install

```sh
agentstack add skill-edouard-claude-pi-flow-skills-flow-help
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# flow-help — orientation

You are a peer facilitator, not a menu. Detect the phase, reason about gates and dependencies, propose ONE single next step with clear rationale.

For the visual sprint dashboard (epics + stories with status symbols), point the user to `/flow-status` — that is a separate, cheap skill. This skill stays light.

## When to use

- Start of a session on a flow-driven project
- User asks 'what do I do next', 'where are we in the process'
- After each flow skill, to decide what comes next

## Inputs

- `.agents/` (whole tree to detect the phase)
- `.agents/implementation/sprint-status.yaml` (if present, for current/next story)
- `git log -10` (recent activity)

## Process

### Step 0 — Cold-start detection (cheap, always done)

If `.agents/memory/overview.md` exists, the project has long-term memory. Check whether the user is **returning after a long gap**:

```bash
last_commit_age_days=$(( ( $(date +%s) - $(git log -1 --format=%ct 2>/dev/null || date +%s) ) / 86400 ))
```

- `last_commit_age_days >= 30` → **cold-start mode**: prefix your output with a Welcome-back block (see Output section). Read:
  - The latest `## État actuel — ` block of `.agents/memory/overview.md` (the last one — earlier ones are archived as `## État au `)
  - The last 3-5 entries of `.agents/memory/journal.md` (tail)
- `last_commit_age_days ` |
| sprint with `review` story (no Senior Review) | — | implementation | `/flow-review ` |
| sprint with `review` story + Senior Review | — | implementation | `/flow-commit ` |
| sprint with `ready-for-dev` / `backlog` stories | — | implementation | `/flow-story ` or `/flow-auto` |
| all stories of an epic `done`, retrospective `optional` | — | end of epic | `/flow-retro` |
| user signals a major change | — | anytime | `/flow-course-correct` |

### Step 2 — Gates and prioritization

- Never recommend a phase without its prerequisite (no `flow-architecture` without PRD, no `flow-story` without sprint-status).
- Prefer `/flow-auto` when several stories are eligible (batch mode).
- If a story is in progress, recommend finishing its cycle before starting a new one.

### Step 3 — Read sprint-status briefly

If `sprint-status.yaml` exists, scan `development_status` to find:
- Currently `in-progress` or `review` story (resume target)
- Next eligible story (the first `backlog` / `ready-for-dev` whose dependencies in `dependencies:` are all `done`)

## Output

### Warm-start (default)

4 lines max:

```
State : 
Reco  : /flow- [args]
Why   : 
Output: 
```

### Cold-start (gap ≥ 30 days, memory layer present)

Prefix the 4-line block with a Welcome-back panel:

```
Welcome back — last activity  days ago.

Project state (per .agents/memory/overview.md):

Recent epics:
-  — : 
-  — : 
-  — : 

State : 
Reco  : /flow- [args]
Why   : 
Output: 
```

Cite verbatim from memory files — never invent. If `overview.md` exists but is empty (no État actuel block yet), say "Project state: memory layer present but no overview yet — run /flow-retro after the next epic."

If the user wants the visual dashboard, append at the end:
```
For the full sprint dashboard, run /flow-status.
```

**Do not launch the next skill.** Propose, the user invokes.

## `.agents/` reference layout

```
.agents/
├── planning/
│   ├── brainstorm-.md
│   ├── product-brief.md
│   ├── current-state.md
│   ├── prd.md
│   ├── architecture.md
│   └── epics/
│       └── epic-XXX.md
├── implementation/
│   ├── sprint-status.yaml   # development_status + dependencies (BMAD-style)
│   ├── stories/
│   │   └── story-XXX.md
│   └── retro-epic-XXX.md
├── memory/                  # long-term, populated by /flow-retro (v0.5+)
│   ├── overview.md
│   ├── decisions.md
│   ├── lessons.md
│   ├── journal.md
│   └── glossary.md
├── internal/                # transient sub-agent outputs (v0.4+)
│   └── /
└── project-context.md
```

## Source & license

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

- **Author:** [edouard-claude](https://github.com/edouard-claude)
- **Source:** [edouard-claude/pi-flow-skills](https://github.com/edouard-claude/pi-flow-skills)
- **License:** MIT

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

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-edouard-claude-pi-flow-skills-flow-help
- Seller: https://agentstack.voostack.com/s/edouard-claude
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.