CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

unified-taskflow

面向真正复杂任务的重型管理系统,使用 anchor、checkpoint、design、恢复和验收追踪。 仅在用户明确要求使用 unified-taskflow、taskflow、.taskflow,或任务确实需要跨多轮、多文件、多阶段的可恢复记录时使用。 不用于小修、普通优化、普通重构、单个 Skill 小改、日常 bug fix,或 task-mode-router / minimal-implementation 已足够处理的任务。

Quality

90%

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

The canonical home for this skill is unified-taskflow in ooooooooooooooooooop/agent-tools

SKILL.md
Quality
Evals
Security

Quality

Content

85%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.

This is a well-engineered process skill: a clearly gated workflow with explicit validation loops, executable lifecycle commands, and a properly layered bundle where every reference and template exists and is well-signaled. The main costs are rule repetition across sections (description rules restated in 触发判断, workflow rules restated in the anti-patterns table) and reliance on references for the concrete artifact formats.

Suggestions

Trim the duplication between the 按需加载 list and the 引用文件 table — merge them into one table (file | purpose | when to load) to save tokens without losing navigation.

Consolidate the repeated '仅目标级/不可逆假设需用户确认' rule into a single stated location (e.g., governance.md or one section) and cross-reference it, rather than restating it in Phase 0, RIPER rules, and the anti-patterns table.

Include a minimal filled-in example of an anchor.md entry (or an excerpt of the template) inline so Phase 0's output format is executable from the body without first reading the template.

DimensionReasoningScore

Conciseness

The body is dense and information-forward with no beginner-concept filler, and the anti-pattern table's duplication is explicitly justified ("禁止性指令的遵从率通常高于义务性指令"). It is not a 5 because several rules are repeated nearly verbatim in multiple sections (e.g., "仅目标级/不可逆假设需用户确认" appears in Phase 0, RIPER rules, and the anti-patterns table; the 续接 semantics appear twice), and the 触发判断 section restates the frontmatter description.

4 / 5

Actionability

Concrete, executable guidance is present: copy-paste commands ("python3 scripts/task-lifecycle.py new demo-task", "session-catchup.py --help"), an explicit .taskflow/ directory layout, a defined checkpoint event table, and named artifacts (anchor.md, checkpoint.md, design.md). It is not a 5 because much of the workflow guidance is procedural rules whose exact file formats live in references/templates, so an agent cannot fully execute Phase 0 outputs from the body alone.

4 / 5

Workflow Clarity

The sequence Phase 0 → 弹性执行 → 完成 is clearly ordered with explicit validation checkpoints: the 完备性门禁 before writing anchor.md, event-driven checkpoint updates with re-grounding 逐项核对, a final Done-when checklist before completion, and a 3-Strike → DEADBAND feedback loop for error recovery. A concrete minimal verification command set is also included.

5 / 5

Progressive Disclosure

The bundle structure matches the body exactly: all four referenced files (phase0-clarification.md, interaction-design.md, governance.md, regrounding-protocol.md) and all three templates under assets/templates/ exist, references are one level deep, and both the 按需加载 section and the 引用文件 table clearly signal what each file is for and when to load it.

5 / 5

Total

18

/

20

Passed

Description

95%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.

The description is strong: it states what the skill does with concrete mechanism names, gives explicit trigger terms (including the .taskflow directory form), and adds explicit exclusions that demarcate it from related routing skills. The only soft spot is that the 'what' leans on internal artifact jargon rather than user-visible outcomes.

DimensionReasoningScore

Specificity

The description lists several concrete mechanisms — "使用 anchor、checkpoint、design、恢复和验收追踪" — giving a specific picture of what the system does. It stops short of a 5 because "重型管理系统" frames capability abstractly and the mechanisms are internal artifacts rather than user-facing actions.

4 / 5

Completeness

Both questions are explicitly answered: what it does ("重型管理系统,使用 anchor、checkpoint、design、恢复和验收追踪") and when to use it ("仅在用户明确要求…或任务确实需要跨多轮、多文件、多阶段的可恢复记录时使用"), with an additional explicit when-not clause. This matches the anchor-5 example structure of what + concrete trigger phrases.

5 / 5

Trigger Term Quality

Comprehensive natural trigger coverage including synonyms and the directory form: "仅在用户明确要求使用 unified-taskflow、taskflow、.taskflow", plus situational phrases like "跨多轮、多文件、多阶段的可恢复记录" that a user would naturally say. It also covers counter-triggers ("小修、普通优化、普通重构…日常 bug fix").

5 / 5

Distinctiveness Conflict Risk

A clear niche with distinct triggers: it names the skill's own invocation terms and explicitly demarcates against neighboring skills ("task-mode-router / minimal-implementation 已足够处理的任务", and "clarify-before-change" in the body). The dual positive/negative trigger list makes wrong-skill triggering unlikely.

5 / 5

Total

19

/

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

relative_links

Relative link issues: 3 deeper-than-1-level

Warning

referenced_paths_exist

Referenced path issues: 3 deeper-than-1-level

Warning

Total

14

/

16

Passed

Repository

ooooooooooooooooooop/personal-ai

Path

skills/unified-taskflow/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.