Repeatable agentic engineering
Raise the quality bar for AI-assisted software work.
Flow-Next is the workflow layer for agentic engineering: durable specs, context-fit planning, re-anchored worker agents, adversarial review loops, docs drift prevention, and receipts for every serious handoff. When you're ready, hand the whole pipeline to a loop — pilot ticks in your session, Ralph runs overnight — with the same quality bar across any platform.
$ /flow-next:plan fn-52 && /flow-next:work fn-52 SHIP fn-52-quality-gates
ADVANCED fn-61 · pilot
$ /flow-next:plan fn-52
✓ sized 4 tasks for focused context
✓ dependencies recorded
$ /flow-next:work fn-52
> Re-anchoring spec + task
+ Running worker subagent
+ Updating flowctl state
+ Evidence recorded
$ /flow-next:impl-review fn-52
✓ verdict: SHIP · codex:gpt-5.5:high
Receipt: .flow/review-receipts/fn-52.json
$ /flow-next:qa fn-52
✓ verdict: SHIP · live-app · 6/6 R-IDs
$ /flow-next:make-pr fn-52
✓ PR opened — R-ID coverage + critical changes
$ ▌Turn vague requests into durable specs that keep product intent, technical constraints, and review criteria stable.
spec: fn-52-quality-gates
source: .flow/specs/fn-52-*.md
R1: Every task maps to acceptance criteria.
R2: Review receipts gate handoff.
R3: PR body tells reviewers where to look.Automatically split specs into dependency-ordered task slices sized for focused agent context windows.
/flow-next:plan fn-52
ready:
fn-52.1 wire config schema
fn-52.2 add review receipt gate
blocked:
fn-52.3 docs sync after API landsRun each task in fresh execution context, rereading the spec, task, git state, and relevant repo code before editing.
/flow-next:work fn-52
> read .flow/specs/fn-52.md
> read ready task + git status
> inspect relevant code
> edit, test, record evidenceRaise the quality bar with plan review, implementation review, completion review, live-app QA, docs drift checks, and receipts.
/flow-next:plan-review fn-52
/flow-next:impl-review fn-52
/flow-next:spec-completion-review fn-52
/flow-next:qa fn-52 # live-app
verdict: NEEDS_WORK -> fix -> review -> SHIPStandardize agent workflow across Claude Code, OpenAI Codex, Factory Droid, xAI Grok Build, Cursor, and review backends.
Claude Code /flow-next:work fn-52
OpenAI Codex /flow-next:work fn-52
Factory Droid /flow-next:work fn-52
Grok Build /flow-next:work fn-52
Cursor /flow-next:work fn-52
RepoPrompt reviews when availableProject specs to Linear or GitHub Issues, reconcile body, status, and comments two-way, and review PRs as Linear Diffs. Projection, not coordination.
/flow-next:tracker-sync
> push fn-52 -> WOR-17 (linked)
> reconcile body / status / comments
> make-pr links PR -> Linear DiffRender specs and PRs as self-contained HTML lenses — a spec visualizer with task DAG and R-ID coverage, and a read-only PR review instrument. Markdown stays the record.
flowctl config set artifacts.html.enabled true
.flow/artifacts/fn-52/
spec.html # thesis, task DAG, R-ID matrix
pr.html # churn map, evidence, checklist01 / Loops
One tick advances one ready spec by one stage — plan, review, work, draft PR — and reports a verdict your host's loop can act on.
/loop 10m /flow-next:pilot Babysits the draft PRs the build loop opened: CI green, reviews converged, merged, released — per your project's own rules.
/loop 30m /flow-next:land External shell loop for fully planned specs: fresh session per iteration, hook-enforced guardrails, receipts on disk. When the run outlasts a session.
./scripts/ralph/ralph.sh tick 1 fn-61 · stage=plan ✓ 4 tasks, deps recorded PILOT_VERDICT=ADVANCED spec=fn-61 stage=plan tick 2 fn-61 · stage=plan-review ✓ verdict SHIP (round 1) PILOT_VERDICT=ADVANCED spec=fn-61 stage=plan-review tick 3 fn-61 · stage=work ✓ 4/4 tasks done · impl-review SHIP PILOT_VERDICT=ADVANCED spec=fn-61 stage=work tick 4 fn-61 · stage=make-pr ✓ draft PR #172 opened (gh-confirmed) PILOT_VERDICT=ADVANCED spec=fn-61 stage=make-pr tick 5 PILOT_VERDICT=NO_WORK spec=- stage=- backlog drained — bless more on the board
02 / Signal
“Flow-next is simply the best coding flow, not even close, and still a side project.”
“Cross-model review is genius. Different models make different mistakes, so using them as mutual reviewers creates a safety net single-model workflows can’t match.”
“I’ve found it generating production-quality code. Far, far better than any of the other tools I’ve tried.”
“As a designer, flow-next finally lets me ship features with confidence. The review loop catches what I miss.”
“The re-anchoring is the quiet superpower. After a long session the agent still knows exactly what it’s building.”
“Ralph mode at night, PRs in the morning. Zero drama. The receipts mean I trust what landed.”
“A force multiplier. Plan once, then watch a team of scouts and reviewers do their jobs.”