Intent Framed Agent
Install
CODEBLOCK0
CODEBLOCK1
Purpose
This skill turns implicit intent into an explicit, trackable artifact at the
moment execution starts. It creates a lightweight intent contract, watches for
scope drift while work is in progress, and closes each intent with a short
resolution record.
Scope (Important)
Use this skill for coding tasks only. It is designed for implementation work
that changes executable code.
Do not use it for general-agent activities such as:
- - broad research
- planning-only conversations
- documentation-only work
- operational/admin tasks with no coding implementation
For trivial edits (for example, simple renames or typo fixes), skip the full
intent frame.
Trigger
Activate at the planning-to-execution transition for non-trivial coding work.
Common cues:
- - User says: "go ahead", "implement this", "let's start building"
- Agent is about to move from discussion into code changes
Workflow
Phase 1: Intent Capture
At execution start, emit:
CODEBLOCK2
Rules:
- - Keep each field to 1-2 sentences.
- Ask for confirmation before coding:
-
Does this capture what we are doing? Anything to adjust before I start?
- - Do not proceed until the user confirms or adjusts.
Phase 2: Intent Monitor
During execution, monitor for drift at natural boundaries:
- - before touching a new area/file
- before starting a new logical work unit
- when current action feels tangential
Drift examples:
- - work outside stated scope
- approach changes with no explicit pivot
- new features/refactors outside constraints
- solving a different problem than the stated outcome
When detected, emit:
CODEBLOCK3
If pivot is intentional, update the active intent frame and continue. If not,
return to the original scope.
Phase 3: Intent Resolution
When work under the active intent ends, emit:
CODEBLOCK4
Resolution is preferred but optional if the session ends abruptly.
Multi-Intent Sessions
One session can contain multiple intent frames.
Rules:
- 1. Resolve current intent before opening the next.
- If user changes direction mid-task, resolve current intent as
Abandoned or
Pivoted, then open a new frame.
- 3. Drift checks always target the currently active frame.
- Number frames sequentially within the session (
#1, #2, ...). - Constraints do not carry forward unless explicitly restated.
Entire CLI Integration
Entire CLI: https://github.com/entireio/cli
When tool access is available, detect Entire at activation:
CODEBLOCK5
- - If it succeeds, mention that intent records will be captured in the session
transcript on the checkpoint branch.
- - If unavailable/failing, continue silently. Do not block execution and do not
nag about installation.
Copilot/chat fallback:
- - If command execution is unavailable, skip detection and continue with the
same intent workflow in chat output.
Guardrails
- - Keep it lightweight; avoid long prose.
- Do not over-trigger on trivial tasks.
- Do not interrupt on every small step.
- Treat acknowledged pivots as valid.
- Preserve exact structured block headers/fields for parseability.
Interoperability with Other Skills
Use this skill as the front-door alignment layer for non-trivial coding work:
- 1.
plan-interview (optional, for requirement shaping) - INLINECODE6 (execution contract + drift monitoring)
- Implementation
- INLINECODE7 (post-completion quality/security pass)
- INLINECODE8 (capture recurring patterns and promote durable rules)
This ordering helps reduce scope drift early and improve repeatability across
tasks.
意图框架代理
安装
bash
npx skills add pskoett/pskoett-ai-skills
bash
npx skills add pskoett/pskoett-ai-skills/intent-framed-agent
目的
该技能在执行开始时将隐性意图转化为显式、可追踪的产物。它创建一个轻量级的意图契约,在工作进行中监控范围漂移,并以简短的结果记录关闭每个意图。
适用范围(重要)
仅将此技能用于编码任务。它专为更改可执行代码的实现工作而设计。
请勿将其用于通用代理活动,例如:
- - 广泛研究
- 仅规划性对话
- 仅文档工作
- 无编码实现的操作/管理任务
对于琐碎的编辑(例如简单的重命名或拼写错误修复),跳过完整的意图框架。
触发条件
在非琐碎编码工作的规划到执行过渡阶段激活。
常见提示:
- - 用户说:开始吧、实现这个、我们开始构建吧
- 代理即将从讨论转向代码更改
工作流程
阶段一:意图捕获
在执行开始时,输出:
markdown
意图框架 #N
成果: [一句话。完成状态是什么样的?]
方法: [我们将如何实现。关键决策。]
约束: [范围外的边界。]
成功标准: [我们如何验证完成。]
预估复杂度: [小 / 中 / 大]
规则:
- 这捕获了我们正在做的事情吗?在我开始之前有什么需要调整的吗?
阶段二:意图监控
在执行期间,在自然边界处监控漂移:
- - 在触及新区域/文件之前
- 在开始新的逻辑工作单元之前
- 当当前操作感觉偏离主题时
漂移示例:
- - 在声明范围之外工作
- 方法变更但未明确转向
- 超出约束的新功能/重构
- 解决与声明成果不同的问题
检测到时,输出:
markdown
意图检查 #N
这看起来可能超出了声明的意图范围。
声明成果: [来自当前框架]
当前操作: [正在发生的事情]
问题: 这是有意的转向还是意外的范围蔓延?
如果转向是有意的,更新当前意图框架并继续。如果不是,则回到原始范围。
阶段三:意图解决
当当前意图下的工作结束时,输出:
markdown
意图解决 #N
成果: [已完成 / 部分完成 / 已转向 / 已放弃]
交付内容: [简要的实际输出]
转向: [任何已确认的变更,或无]
待办事项: [剩余范围内事项,或无]
解决是首选,但如果会话突然结束则为可选。
多意图会话
一个会话可以包含多个意图框架。
规则:
- 1. 在打开下一个意图之前解决当前意图。
- 如果用户在任务中途改变方向,将当前意图解决为已放弃或已转向,然后打开新框架。
- 漂移检查始终针对当前活动框架。
- 在会话内按顺序编号框架(#1、#2、...)。
- 除非明确重申,否则约束不会延续。
Entire CLI 集成
Entire CLI:https://github.com/entireio/cli
当工具可用时,在激活时检测 Entire:
bash
entire status 2>/dev/null
- - 如果成功,说明意图记录将被捕获在检查点分支的会话记录中。
- 如果不可用/失败,静默继续。不要阻塞执行,也不要纠缠安装事宜。
Copilot/聊天回退:
- - 如果命令执行不可用,跳过检测并在聊天输出中继续相同的意图工作流程。
防护措施
- - 保持轻量级;避免冗长的叙述。
- 不要对琐碎任务过度触发。
- 不要在每个小步骤上中断。
- 将已确认的转向视为有效。
- 保留精确的结构化块头/字段以便解析。
与其他技能的互操作性
将此技能作为非琐碎编码工作的前端对齐层:
- 1. plan-interview(可选,用于需求塑造)
- intent-framed-agent(执行契约 + 漂移监控)
- 实现
- simplify-and-harden(完成后的质量/安全检查)
- self-improvement(捕获重复模式并推广持久规则)
此顺序有助于早期减少范围漂移并提高跨任务的可重复性。