# Ase Task Condense > > - **Type:** Skill - **Install:** `agentstack add skill-rse-ase-ase-task-condense` - **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-task-condense - **Website:** https://ase.tools ## Install ```sh agentstack add skill-rse-ase-ase-task-condense ``` 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 Condense a Task Plan $ARGUMENTS *Condense* the task plan by removing fluff while preserving all semantics exactly. @${CLAUDE_SKILL_DIR}/../../meta/ase-format-task.md Procedure --------- 1. **Determine Task:** 1. Set initially. Inherit the always existing from the current context. Inherit the always existing from the current context. Do not output anything. 2. React on task id: 1. matches the regexp `^[a-zA-Z][a-zA-Z0-9_-]*$` "> Set (set task id to instruction) and (set instruction 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. is NOT empty"> The argument is neither empty nor a valid task id. As this skill only accepts an optional `[]` argument and *never* a free-text instruction, only output the following and then immediately *STOP* processing the entire current skill: ⧉ **ASE**: ☻ skill: **ase-task-condense**, ▶ ERROR: expected single `[]` argument 2. **Determine Operation:** 1. Call the `ase_task_load(id: "")` tool of the `ase` MCP server to load the current task plan content and set to the `text` output field of the `ase_task_load` tool call. - If starts with `ERROR:` or `WARNING:`: Set (set content to empty). Set to "0". - If starts NOT with `ERROR:` and NOT with `WARNING:`: Set (set content to text). Calculate the number of words of . Set (remember the loaded word count for the strictly-smaller check in step 3). contains '⎈ Created: '"> Set (extract the original creation timestamp so it can be re-inserted unchanged into the condensed in step 3). Only output the following : ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan loaded** 2. is empty"> Complain and tell the user to use the `ase-code-resolve`, `ase-code-refactor`, `ase-code-craft`, or `ase-task-edit` skills first to create a task plan. Then immediately stop processing this skill. 3. **Condense Task Plan:** 1. *Apply the condense ruleset* to , producing a shorter . The goal is to make the plan require as *little reading* as possible while all semantics remain *fully preserved and unchanged*. Honor the following ruleset *strictly*: 1. *Preserve-exactly (never alter)*: the plan structure (the headings `#`/`##`, all three `## CONTEXT`, `## CHANGES`, and `## VERIFICATION` sections, the `Created:`/`Modified:` lines, and the `- ****:` bullet labels), all *code spans* and code blocks, technical terms, file paths, identifiers, numbers, severities (`LOW`/`MEDIUM`/`HIGH`/`ACCEPTED`), and the `**` emphasis highlighting convention. 2. *Compress free-text only* (the `**WHAT**`/`**WHY**` prose and each bullet's `` text): - *Drop* filler ("just", "really", "basically", "simply"), pleasantries, and hedging ("I think", "maybe", "perhaps"). - *Use* shorter synonyms and common abbreviations. - *Use* `→` for causality and `-` for short subsequent facts. - *Drop* articles ("a", "an", "the") and *replace* conjunctions with short separate clauses where this shortens the text without introducing ambiguity. - *Re-wrap* the shortened free-text to the ~120-character- per-line convention. - *Merge* genuinely-redundant bullets (the same aspect restated) and *drop* pure duplication - but *only* when truly redundant; *never* lose a distinct aspect. 3. *Persona override*: this condense ruleset *always wins* for the plan content. This ruleset-based compression is applied *regardless* of the currently active session persona style. 4. *Hard guardrail - semantics preserved EXACTLY*: condensing *only* shortens wording. It *MUST NOT* drop, merge (except truly-redundant bullets per sub-item 2), reorder, or alter *any* factual claim, requirement, file path, rule, or example. If a shortening would change meaning, *keep the longer wording*. 2. *Persist only if smaller*: calculate the number of words of the condensed . - is strictly smaller than "> Update with the current time in ISO-style format, which has to be determined by calling the `ase_timestamp(format: "yyyy-LL-dd HH:mm")` tool of the `ase` MCP server and use the `text` field of its response. Keep the original , re-insert the current and the refreshed into . Call the `ase_task_save(id: "", text: "")` tool of the `ase` MCP server to save the condensed task plan content. Do not output anything related to this MCP call except the following : ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan condensed** - is NOT strictly smaller than "> Do *not* save and do *not* bump the timestamp. Only output the following : ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan already condensed** 4. **Decide Next Step:** 1. *Determine next step*: - If is not equal to `none`: Treat as a comma-separated chronological list of pre-selected next-step tokens. *Split* it on `,`, take the *first* token as , and store the remaining tokens (joined back with `,`, or `none` if empty) into so downstream skills can consume the tail. - If matches the regex `^(DONE|EDIT|IMPLEMENT|PREFLIGHT)$`: Honor the pre-selected token. Set . - else: Only output the following and then immediately *STOP* processing the entire current skill: ⧉ **ASE**: ☻ skill: **ase-task-condense**, ▶ ERROR: invalid `--next` token: **** - If is equal to `none`: 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: Next Step: How would you like to proceed with the plan? DONE: Stop processing. EDIT: Hand off plan to editing. IMPLEMENT: Hand off plan to implementation. PREFLIGHT: Hand off plan to pre-flighting. 2. Check the tool and dispatch accordingly: - If is `DONE` or `CANCEL`: Only output the following and then *STOP*. ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan condensed -- done** - If is `EDIT`: Set (set args to empty). Set --int-reuse-task. is not equal `none`"> Set --next Only output the following and then call the tool `Skill(skill: "ase:ase-task-edit", args: "")` to invoke the `ase:ase-task-edit` skill in order to *edit* the condensed plan. Immediately stop processing the current skill once the `Skill` tool was used. ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan condensed -- hand-off to edit** - If is `IMPLEMENT`: Set (set args to empty). Set --int-reuse-task. is not equal `none`"> Set --next Only output the following and then call the `Skill(skill: "ase:ase-task-implement", args: "")` tool to *apply* the plan. ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan condensed -- hand-off to implementation** - If is `PREFLIGHT`: Set (set args to empty). Set --int-reuse-task. is not equal `none`"> Set --next Only output the following and then call the `Skill(skill: "ase:ase-task-preflight", args: "")` tool to *apply* the plan. ⧉ **ASE**: ◉ task: ****, ✪ plan: **** words, ▶ status: **plan condensed -- hand-off to pre-flight** ## 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-task-condense - 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%.