flow-plan
Skill0 starsChanged 3 months agoApache-2.0View on GitHub
Flow phases 3–4 (planning + gated plan review). Normally invoked via /flow. Writes a lean plan.md, runs the plan-critic, then holds at gate 1 for user approval.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- baodhg/FLOWtheir own version, first seen Nov 8, 2026
- emiliodominguez/flow-skillstheir own version, first seen Oct 15, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- baodhg/FLOW★ 2same name
- emiliodominguez/flow-skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- flow-implementSkill
- flow-reportSkill
- flow-researchSkill
- flow-reviewSkill
- flow-shapeSkill
- flowSkill
- flow-startSkill
- flow-wrapSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 16
Write plan.md in the task folder from the plugin's templates/plan.md:
≈ “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.
Files
Flow — plan & plan review
Phase 3: write the plan (inline, user watching)
Lite mode only — triage unknowns first. Lite skips research, so this is
the only place they get caught. Sort them as flow-research step 4 does and
ask the shape-changing ones in ONE AskUserQuestion round before writing
anything; log the answers to the decision log. Needing a second round means
lite was the wrong mode — say so.
Write plan.md in the task folder from the plugin's templates/plan.md:
- Intent + falsifiable acceptance criteria (each checkable by a command, test, or observation).
- Approach chosen, alternatives noted in one line each.
- Assumptions & open questions — every unknown resolved without asking,
marked
[answered]/[assumed]/[open — owner]. An empty section claims the plan made zero judgment calls; that is rare, not the default. - Tasks: file-level — every task names the files it touches and its verification step. Guardrail: if the Tasks section is longer than the diff you expect it to produce, cut it down before proceeding.
- Out of scope + Verification (machine commands AND how to drive the app/feature end-to-end).
Then set phase: plan-review, update Next action.
Phase 4: gated review
- Critic pass (once). Dispatch
flow-plan-criticwith the plan.md, research.md, and repo paths. Revise plan.md for Blocking findings; use judgment on Advisory ones. ONE critic pass — do not re-dispatch after revising. - User gate. Present: plan.md by path, a ≤5-line summary of the approach, one line tallying the Assumptions section ("3 assumptions — 1 you answered, 1 repo default, 1 left open"), and the critic's verdict summary. Ask approve / revise (AskUserQuestion). The tally is disclosure, not a fresh question round — the asking already happened upstream; this only makes sure approval is informed. Revisions loop back to the user, not to the critic.
- On approval (never before): append to the decision log
- YYYY-MM-DD gate 1: plan approved by user, setphase: implementation, updateupdated+Next action, announce, and invoke theflow-implementskill.
The validator blocks phase: implementation without the gate-1 line and
without a filled Assumptions section in plan.md, and blocks all code edits
until then. A plan that genuinely made no judgment calls writes
- none — <reason>; an empty section is not a way out. If blocked, you
skipped a step — fix the process, never the guardrail.
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
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.

