Install
$ agentstack add skill-aldianriski-lean-flow-flow ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo 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.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
flow
The opt-in conductor for the lean-flow loop. It drives the standalone skills through the proper sequence so the discipline runs instead of being remembered — à la carte still works; this is the conducted path. It sequences; it never bypasses a gate.
> Every skill stays standalone (nothing requires another). /flow is the one component allowed to > depend on the others — conducting them is its whole job. It does not re-implement any stage; it calls them.
What it conducts
Assess the current state first, then run only the stages whose precondition is unmet — state which you're running and why you're skipping the rest:
- Orient — context missing/stale →
/prime. Resuming from a handoff? read it first. - Feed — no open work →
/task-decomposer ""— the detailed grill fires here, at intake (the build gates re-grill only residuals); backlog drifted / noisy →/triage. - Plan — no active sprint but a
readybacklog exists →/lean-doc-generator promote(the governance review fires here). - Build —
/orchestrator sprint-bulkthrough G1 / G2. Route each task by type: new behaviour →/tdd· bug →/diagnose· hard-to-change →/refactor-advisor· a design that must be felt →/prototypefirst. - Close — all DoD ticked →
/lean-doc-generator close(Retro → §10 buckets), then/release-patch(PATCH) for a fixes-only sprint · MINOR by hand for a feature sprint. - Continuity — stopping mid-loop or context budget low →
/handoff, so the next/flowresumes cleanly.
Pin the target with /goal — set a /goal to the active task/sprint's DoD so the conducted run keeps driving across turns until it's verifiably met; clear it at close. That's the native engine behind "conduct the loop".
Conductor, not autopilot
- Never self-approve a gate. G1 Scope and G2 Design still need explicit human sign-off —
/flowpauses at each; it does not wave them through. - Halt on first blocker (BLOCKED / CRITICAL / human
block) — report and wait. - One sprint per stream — never a second sprint in a stream that already has one; parallel streams (sprint
stream:frontmatter) each run their own. Single-stream repos: exactly one active sprint, as before. - The conducted path produces the same artifacts as à la carte — it only guarantees the sequence and that the Close governance actually runs.
Red flags
❌ Auto-approving G1/G2 — the conductor sequences; the human gates. ❌ Skipping the Close governance (Retro / §10) because "the build's done" — that's the part that compounds. ❌ Re-running a stage whose output already exists — assess first; skip with a stated reason. ❌ Conducting when one skill was wanted — /flow is opt-in; a single /tdd is not a reason to drive the whole loop. ❌ Re-implementing a stage inline — always call the standalone skill, never inline its logic.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: aldianriski
- Source: aldianriski/lean-flow
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.