frappe-task
Skill0 starsChanged 3 months agoMITView on GitHub
End-to-end Frappe/ERPNext task workflow - plan, implement, review with Codex, commit, deploy, and hand off to testing. User-invoked only.
In this repository8 other files it gives its agents.
- codex-reviewSkill
- deploymentSkill
- feature-changelogSkill
- git-finalizationSkill
- project-contextSkill
- task-implementationSkill
- task-planningSkill
- testing-taskSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Keep this file's rules loaded.
- Very specific: line 33
All deterministic facts come from the helper CLI at ${CLAUDE_PLUGIN_ROOT}/bin/frappe-workflow (detection, state, gates, fingerprints, IDs, …
- Very specific: line 40
Every shared workflow file lives under docs/ai-context/ and is Git-trackable, so an active task can continue on another computer after the …
- Distinctive: line 49
Never commit, push, open SSH, or deploy without the explicit user confirmation each of those steps requires.
- Very specific: line 83
When invoked with no action and state exists: validate it (bin/frappe-workflow validate workflow-state), cross-check against Git, then …
≈ “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
Frappe Task Workflow
You are the orchestrator of a staged Frappe development workflow. You route actions, enforce gates, and delegate detailed work to the internal skills. Keep this file's rules loaded; read the referenced files only when the current action needs them (progressive disclosure).
Parse the Arguments
$ARGUMENTS — the first whitespace-separated token is the action; everything
after it is the action input.
Recognized actions: init, start, status, review, apply-review,
commit, deploy, testing, reset, help.
- No arguments → resume the active task from its persisted workflow
stage, recorded in
docs/ai-context/task-workflow.json. This is how work continues after Claude Code is closed and reopened. - First token not a recognized action → see "Unknown input" in references/command-routing.md.
Ground Rules (always)
- Read ../../references/shared-workflow-rules.md before acting on any workflow-mutating action.
- All deterministic facts come from the helper CLI at
${CLAUDE_PLUGIN_ROOT}/bin/frappe-workflow(detection, state, gates, fingerprints, IDs, scans). Never guess what it can compute. - The workflow state file
docs/ai-context/task-workflow.jsonin the target app repository is the primary logical state; Git verifies it. Stage changes go only throughbin/frappe-workflow state transition <stage>. - Every shared workflow file lives under
docs/ai-context/and is Git-trackable, so an active task can continue on another computer after the user commits and pushes the working branch. Only.claude/deployment.local.jsonand.claude/task-workflow.lockstay machine-local. Never commit, push, or pull automatically. - The nine canonical stages are
planning,implementation,codex_review,review_fixes,ready_for_commit,committed,deployment_skipped,deployed,completed. Never name a stage that is not on this list, in output or in state. - Never commit, push, open SSH, or deploy without the explicit user confirmation each of those steps requires.
Routing
Full per-action procedures: references/command-routing.md. Stage semantics: references/workflow-stages.md. Allowed transitions: references/state-transitions.md. Gates: references/workflow-gates.md.
What each action does, and which skill owns it. These descriptions are
user-facing and must stay accurate — help prints them (verbatim, from
examples/help-output.md), so a wrong
description here becomes a wrong answer to the user.
| Action | What it does | Delegate to |
| (none) | Resume the active task from its persisted stage | — (see command-routing.md §No Action) |
init | Initialize the application: detect bench/app/Sites, generate or validate docs/ai-context/PROJECT_CONTEXT.md and docs/ai-context/FEATURE_CHANGELOG.md, migrate a legacy layout, and prepare shared workflow storage. Never starts a task. | project-context + feature-changelog skills |
start | Accept a prepared plan or a task description; produce a validated, repository-aware docs/ai-context/TASK_PLAN.md | task-planning skill |
status | Read-only report of task, stage, progress, review, commit, deployment, blockers, inconsistencies | — (see examples/status-output.md) |
review | Validate the completion gate and generate a Codex review prompt. Codex is never run automatically. | codex-review skill |
apply-review | Process an APPROVED or CHANGES_REQUIRED result, record the review round, route to review_fixes or ready_for_commit | codex-review skill |
commit | Prepare the Conventional Commit message and exact staging commands; execute only on explicit request | git-finalization skill |
deploy | Ask deploy-or-skip, then run the safe deployment procedure | deployment skill |
testing | Generate a concise Arabic testing-team title and description from the approved behavior and print them in the terminal for copying — no file is created | testing-task skill |
reset | Reset active task workflow state after explicit confirmation | — (see command-routing.md §reset) |
help | Print the canonical help text, read-only | — (see examples/help-output.md) |
During the implementation stage (reached after start, or when resuming
into it), follow the task-implementation skill.
Resuming After Restart
When invoked with no action and state exists: validate it
(bin/frappe-workflow validate workflow-state), cross-check against Git,
then resume work from the recorded current_stage. On inconsistencies, follow
../../references/error-and-blocker-handling.md —
stop and report rather than guessing. When no state exists, tell the
user to run /frappe-workflow:frappe-task init; do not create a task
automatically.
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.