CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

thread-orchestration

大任务的主动拆解、动态角色分配与多 thread 并行编排。 Use when: 任务涉及 2+ 个独立可交付子任务,或需组织指挥、执行、运行后验收之间的交付与通讯。 Not for: 单一任务(直接做)、已有 thread 之间的被动协调(用 cross-thread-sync)、单 session 内 subagent 并行(CLI 内置能力)、发现跨 scope 问题但已有归属 thread(用 cross_post_message,不要新建 thread)。 Output: 角色与依赖计划、获批子 thread 派发、横向交付与最终回流。 GOTCHA: projectPath 是子 thread 的工作区/真相源归属,不是外部目标仓;社区 PR review 目标可以是 clowder-ai,但工作区仍可能应继承 cat-cafe。

Quality

78%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./cat-cafe-skills/thread-orchestration/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

62%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, highly actionable orchestration SOP with an exemplary gated workflow, but it pays for it in token efficiency: the same rules (reportingMode, proposal approval) are repeated across four to five sections and internal changelog/ticket references are embedded inline. The single referenced bundle file is also missing from the skill directory, breaking the one progressive-disclosure pointer it has.

Suggestions

State each core rule once: consolidate the reportingMode semantics into the Step 2 table and have Step 5, Quick Reference, and Common Mistakes reference it in one line each instead of restating the four modes and their meanings verbatim in every section.

Move the community-PR/opensource-ops dispatch details, the projectPath case study, and changelog notes (Phase AA 更新, F128/F193/KD-E4/AC-*/F317 citations) into a separate reference file, keeping only the operative rule in SKILL.md.

Fix the dangling reference: either ship refs/constellation-collaboration.md in the bundle or remove the [星系协作契约] links, since following them currently leads nowhere.

DimensionReasoningScore

Conciseness

Core rules are systematically restated many times: reportingMode semantics appear in the Step 2 table, the 场景化选择指南 bullets, the Step 5 mode preamble, Quick Reference (two near-duplicate lines: "reportingMode = ..." and "汇报 = ..."), and Common Mistakes rows; the "propose ≠ create, wait for approval" rule appears at least five times. Inline changelog metadata ("Phase AA 更新(2026-06-07)", F128/F193, KD-E4, AC-AA2/AC-AA6, F317 source citations) is time-sensitive bookkeeping that adds tokens without aiding execution, matching the 'noticeably verbose with several padded sections' anchor rather than the mostly-efficient anchor 3.

2 / 5

Actionability

Concrete, near copy-paste tool call signatures with field-by-field commentary (`cat_cafe_propose_thread` with title/reason/preferredCats/projectPath/reportingMode), a filled-in `cross_post_message` content template with the mandatory `## 主 Thread` header, naming rules with examples, and decision tables. It stops short of the 5 anchor because values are placeholders with no fully worked end-to-end example, and complete structured-dispatch examples are delegated out to `cross-cat-handoff` — minor gaps, not vague direction.

4 / 5

Workflow Clarity

The five-step flow (拆解 → 提议 → 选猫 → 执行 → 交付) is clearly sequenced with explicit validation checkpoints: user approval before any thread exists, per-mode Step 5 behavior, non-author review gates, 运行后验收, and a blocking-ack commit confirmation step with a "主 thread 可能会要求修改后再 commit" recovery loop. The Common Mistakes table plus Quick Reference act as checklists, matching the top anchor.

5 / 5

Progressive Disclosure

The body links one clearly-signaled, one-level-deep reference ([星系协作契约](refs/constellation-collaboration.md)), but no refs/ directory or bundle file exists at that path — the link dangles. Substantial tangential content is also inlined that belongs in reference files (the opensource-ops community-PR dispatch block, reportingMode constraint details), fitting the 'some structure, references present, content that should be separate is inline' anchor; it is above anchor 2 because headers, tables, and the single external pointer are well organized and clearly signaled.

3 / 5

Total

14

/

20

Passed

Description

87%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description: explicit what/when/output structure, concrete trigger conditions, and unusually good boundary setting via the Not-for clause against adjacent skills. The main weakness is dense internal jargon that slightly obscures the concrete actions for a general reader.

DimensionReasoningScore

Specificity

The description lists several concrete actions — "大任务的主动拆解、动态角色分配与多 thread 并行编排" — plus an explicit Output clause naming deliverables ("角色与依赖计划、获批子 thread 派发、横向交付与最终回流"). It falls short of the 5 anchor because the phrasing is dense domain jargon (e.g., "横向交付与最终回流") rather than fully concrete, user-parseable actions, but is well above naming the domain with generic verbs.

4 / 5

Completeness

It explicitly answers both questions: what ("主动拆解、动态角色分配与多 thread 并行编排" with a concrete Output list) and when ("Use when: 任务涉及 2+ 个独立可交付子任务,或需组织指挥、执行、运行后验收之间的交付与通讯"), with concrete trigger conditions and an explicit Not-for exclusion list. Matches the 5 anchor's what+when-with-concrete-triggers pattern; nothing is only implied.

5 / 5

Trigger Term Quality

The description field itself contains natural, sayable terms — "拆解", "多 thread 并行编排", "2+ 个独立可交付子任务", "并行" — that a user would plausibly utter when needing this skill. A few natural variants (e.g., "分 thread", "并行推进", "任务分解") appear only in the separate `triggers` frontmatter field rather than the description, so coverage is good but not comprehensive within the evaluated text.

4 / 5

Distinctiveness Conflict Risk

The "Not for" clause explicitly disambiguates the four nearest neighbors — single tasks (do directly), passive cross-thread coordination (cross-thread-sync), in-session subagent parallelism (CLI built-in), and cross-scope issues with an owning thread (cross_post_message). This is a clear niche with minimal wrong-skill trigger risk.

5 / 5

Total

18

/

20

Passed

Validation

87%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

relative_links

Relative link issues: 2 missing

Warning

Total

14

/

16

Passed

Repository

zts212653/clowder-ai

Path

cat-cafe-skills/thread-orchestration/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.