# Ase Code Refactor > > - **Type:** Skill - **Install:** `agentstack add skill-rse-ase-ase-code-refactor` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [rse](https://agentstack.voostack.com/s/rse) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** Apache-2.0 - **Upstream author:** [rse](https://github.com/rse) - **Source:** https://github.com/rse/ase/tree/master/plugin/skills/ase-code-refactor - **Website:** https://ase.tools ## Install ```sh agentstack add skill-rse-ase-ase-code-refactor ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About @${CLAUDE_SKILL_DIR}/../../meta/ase-control.md @${CLAUDE_SKILL_DIR}/../../meta/ase-skill.md @${CLAUDE_SKILL_DIR}/../../meta/ase-dialog.md @${CLAUDE_SKILL_DIR}/../../meta/ase-getopt.md Refactor Artifacts $ARGUMENTS is equal `true`"> The `--quick`/`-Q` flag is a *shorthand alias*: set to `true`, to `true`, and to `IMPLEMENT,DELETE`. Do not output anything. *Refactor* existing artifacts the following way: @${CLAUDE_SKILL_DIR}/../../meta/ase-format-task.md @${CLAUDE_SKILL_DIR}/../../meta/ase-tenets.md Procedure --------- You *MUST* *NOT* call `Edit`, `Write`, `NotebookEdit`, or any filesystem-modifying tool during this entire skill. The *only* permitted way to persist artifacts is via `ase_task_save(...)`. 1. 1. matches the regexp `^[a-zA-Z][a-zA-Z0-9_-]*$` "> Set (set task id to request) and (set request empty), call the `ase_task_id(id: "", session: "")` tool from the `ase` MCP server to switch the task, and then only output the following : ⧉ **ASE**: ◉ task: ****, ▶ status: **task given** 2. has the format `: ` AND matches the regexp `^[a-zA-Z][a-zA-Z0-9_-]*$` "> Set and and call the `ase_task_id(id: "", session: "")` tool from the `ase` MCP server to implicitly switch the task. Do not output anything. 3. is empty"> Ask the user interactively, without a special tool, for the initial request with a single question: `**No refactoring details known yet. What is the refactoring you want to request?**` Then set to the response of the user. 4. is equal `default` and is not empty "> Set to a unique task id, derived from , which consists of two lower-case words concatenated with a `-` character. Then call the `ase_task_id(id: "", session: "")` tool from the `ase` MCP server to implicitly switch the task. Do not output anything. 5. Report the task and request with the following : ⧉ **ASE**: ◉ task: **** ⧉ **ASE**: ⇌ request: **** 6. Figure out what the artifact refactoring is about. 7. Ask the user for clarification if the goal of this refactoring is too unclear. 8. Do not output anything in this step, unless you asked the user. 2. 1. Check the existing source files for all code which is related to the refactoring . 2. Check the architecture of the existing code base to understand the overall structures and dynamics. 3. Do not output anything in this STEP 2. 3. 1. You *MUST* internalize and strictly honor the **GENERIC TENETS**, and the **REFACTORING TENETS** of the **ASE Tenets** when updating in the following. Do not output anything. 2. Do not output anything in this STEP 3. 4. You *MUST* perform the following sub-steps *internally* and *without any output* until and including the recommendation decision. Only sub-steps 4-6 below are allowed to produce output, and only if is equal `false`. If is equal `true`, *skip* the reporting sub-steps 4-6 entirely (perform no output at all) to speed up processing. 1. *Propose* corresponding *refactoring approach*, including optionally, some *alternative* refactoring approaches. Do *not* output anything in this sub-step. 2. *Reflect* on and *critique* the proposed approaches by deriving, per approach, a small set of concrete *pros* and *cons*. Do *not* output anything in this sub-step. 3. Based on the reflection, *decide* which approach to recommend and annotate it with an of ` ⚝ **RECOMMENDATION** ⚝`. All other approaches receive an empty . Do *not* output anything in this sub-step. 4. Indicate start of reporting by showing the following : 5. Now report each approach with the following , inlining its pros/cons derived in sub-step 2, and do not output anything else in this step: ● **APPROACH A**: **** ○ [...] ⊕ *PRO*: [...] ⊖ *CON*: [...] Hints: - Give a short one-sentence of the refactoring approach plus *precise* and *ultra brief and concise* refactoring information. Try to keep the number of bullet points (`○ [...]`) in the range of 1-4. 6. Indicate end of reporting by showing the following : 7. is not `true`"> In the following, you *MUST* *NOT* use your built-in tool! Instead, you *MUST* just show a custom dialog according to the expanded `custom-dialog` definition. You *MUST* closely follow this definition. Let the user choose the preferred approach A by raising a question with the following custom dialog, where per approach A, you determine an ultra brief summary and then use the answer option `A: ⚝ **RECOMMENDATION** ⚝ - ` for your recommended approach plus zero or more answer options `A: ` for all other approaches: Select Approach: Select your preferred refactoring approach to follow? A: [...] Set to the number of the refactoring approach A you recommend. Output a hint with the following : ⧉ **ASE**: ◉ task: ****, ▶ status: **auto-chosen approach A** 5. 1. *Compose a refactoring plan* for the chosen refactoring A by closely aligning to the existing architecture and the existing code base. Use the defined for a task plan and inject the information from refactoring A and all derived realization decisions into it. Store the resulting task plan in . If a `CHANGELOG.md` file exists in the project (or in any affected sub-package), the plan *MUST* include, as part of its `## ※ CHANGES` section, an explicit bullet point describing the addition of a corresponding new entry to that `CHANGELOG.md` file, aligned with its existing style and conventions. is equal `true`"> You *MUST* completely omit the `## VERIFICATION` section (including its heading and all of its bullet points) from . You *MUST* *NOT* call `Edit`, `Write`, `NotebookEdit`, or any filesystem-modifying tool during this step. 2. Call the `ase_timestamp(format: "yyyy-LL-dd HH:mm")` tool of the `ase` MCP server and use the `text` field of its response for and information. Then insert the current , , and information and calculate the number of words of . 3. You then *MUST* *save* the resulting plan content with the `ase_task_save(id: "", text: "")`. 4. Output a hint with the following : ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan created** 5. Directly pass-through control to the next skill: Treat as a comma-separated chronological list of pre-selected next-step tokens. *Peek* the *first* token as (or `none` if the list is `none`/empty). Set --int-reuse-task. 1. is equal `DONE`"> Consume the head: set to the remaining tokens (joined back with `,`, or `none` if empty). `DONE` means the freshly composed plan is finalized as-is, so do *not* hand off to `ase-task-edit`. Only output the following and then *STOP*. Do *not* implement the plan. ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan finalized -- done** 2. is equal `IMPLEMENT`"> Consume the head: set to the remaining tokens (joined back with `,`, or `none` if empty). is not equal `none`"> Set --next Call the tool `Skill(skill: "ase:ase-task-implement", args: "")` to *implement* the freshly composed plan, bypassing `ase-task-edit`. 3. is equal `PREFLIGHT`"> Consume the head: set to the remaining tokens (joined back with `,`, or `none` if empty). is not equal `none`"> Set --next Call the tool `Skill(skill: "ase:ase-task-preflight", args: "")` to *preflight* the freshly composed plan, bypassing `ase-task-edit`. 4. Hand off to `ase-task-edit`. is equal `EDIT`"> Consume the head: set to the remaining tokens (joined back with `,`, or `none` if empty). `EDIT` is this skill's own dispatch token, *not* part of `ase-task-edit`'s `--next` vocabulary, so it must be stripped here rather than forwarded. All remaining tokens are `ase-task-edit`'s own vocabulary and are forwarded verbatim, so `ase-task-edit` consumes its own head itself. is not equal `none`"> Set --next Then call the tool `Skill(skill: "ase:ase-task-edit", args: "")`. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [rse](https://github.com/rse) - **Source:** [rse/ase](https://github.com/rse/ase) - **License:** Apache-2.0 - **Homepage:** https://ase.tools Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-rse-ase-ase-code-refactor - Seller: https://agentstack.voostack.com/s/rse - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.