* feat(cli): improve agent prompts/skills and workspace fork workflow
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(cli): refuse fork --from-branch rename of a base branch
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(cli): auto-detect fork branch workflow, drop rt.d.ts refresh and legacy-name warning
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(skills): reconcile raw-app generate-metadata stance (agent offers+runs)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(skills): agent runs all CLI commands, gated on intent not on user typing them
Extends #9467's safe-vs-destructive model: the agent runs consequential commands (sync push, generate-metadata) itself too, gated on explicit user intent rather than handed to the user to type. The explicit-intent rule is the safeguard; an approval prompt is treated as a possible backstop, not assumed (auto-approve/headless runs have none).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Revert "docs(skills): agent runs all CLI commands, gated on intent not on user typing them"
Reverts 9225e1759b. That commit over-reached: #9467 already established the safe-vs-destructive split, and the targeted item-6 fix already removed the passive "tell the user they can run <safe next step>" phrasing. The blanket "agent runs everything" principle pushed deploys to be more eager and carried a wrong "permission layer prompts for approval" claim (untrue in auto-approve/headless mode). Keep deploys conservative.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(cli): default fork workspace name/id to the current branch when renaming it
When 'wmill workspace fork' converts the current working branch into the fork branch, default the fork's name and id to that branch (sanitized to a slug, since branch names can contain '/'). Interactive: the prompt is pre-filled (enter to accept); non-interactive (--yes): used automatically. Adds a unit test for the slug derivation.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(cli): address fork review — guard fork-branch rename, cap+validate fork id
Two P2s from review:
- --from-branch refused when the current branch is already a fork branch (would detach the existing fork by renaming its branch).
- fork id slug capped to 42 chars (backend max 50 incl. wm-fork- prefix); auto-derived id is slugged; full id validated client-side before existsWorkspace/datatable cloning so an invalid id fails fast instead of leaving cloned Postgres databases behind.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
System Prompts
This directory contains the single source of truth for AI system prompts used by both the frontend copilot and CLI guidance.
Structure
system_prompts/
├── base/ # Core instruction templates (manually written)
│ ├── flow-base.md # Shared OpenFlow structure guidance
│ └── flow-cli.md # CLI/local-agent workflow guidance for write-flow skill
├── languages/ # Language-specific instructions (manually written)
└── auto-generated/ # Auto-generated files (DO NOT EDIT)
├── sdks/ # SDK documentation
├── cli/ # CLI command documentation
├── prompts.ts # TypeScript exports
└── index.ts # Helper functions
Usage
Regenerating Prompts
When SDK methods or the OpenFlow schema change, run:
python system_prompts/generate.py
To also refresh the standalone skills in a Claude plugin checkout:
python system_prompts/generate.py --plugin-dir ~/windmill-claude-plugin
--plugin-dir accepts:
- the
windmill-claude-pluginrepo root - a plugin root such as
plugins/windmill - a direct
skills/directory
To regenerate the public docs repo (consumed by context7):
python system_prompts/generate.py --context7-dir ~/windmill-cli-docs
--context7-dir writes a fully-rendered snapshot (AGENTS.md,
cli-commands.md, skills/<name>/SKILL.md, README.md, manifest.json
with the Windmill version) with all template placeholders resolved —
suitable for ingestion by docs aggregators. In CI this runs from
.github/workflows/publish-cli-docs.yml on every release tag. The
generator refuses to wipe the target directory unless it's empty or has
a context7 marker (context7.json, manifest.json, or a
windmill-cli-docs git remote), so a typo can't delete unrelated files.
This will:
- Parse TypeScript and Python SDK files to extract function signatures
- Parse the OpenFlow YAML schema
- Parse the CLI commands
- Assemble complete prompts from markdown files
- Generate TypeScript exports in
auto-generated/ - Optionally refresh plugin-ready standalone
SKILL.mdfiles in the target directory
Scope
These system prompts contain ONLY:
- How to write Windmill scripts (language syntax, conventions, SDK usage)
- How to structure Windmill flows (OpenFlow schema, module types, data flow)
- Resource type handling, S3 operations
They DO NOT contain:
- Tool usage instructions (edit_code, set_flow_json, etc.)
- IDE/editor specific commands
- Testing tool invocations
Tool instructions are added separately by the frontend and CLI.
CLI-only workflow instructions live in base/flow-cli.md and are included in the
generated write-flow skill for wmill init. They are intentionally excluded
from the frontend flow chat prompt.
Integration
Frontend
Uses Vite path alias $system_prompts pointing to auto-generated/:
import { FLOW_GUIDANCE } from "$system_prompts/flow";
import { getLangContext } from "$system_prompts/languages";
CLI
Generates /cli/src/guidance/skills.gen.ts with embedded skill content for wmill init.
Editing Guidelines
- Edit markdown files in
base/,languages/ - Never edit files in
auto-generated/directly - After editing, run
generate.pyto update exports