📦 deps(thirdparty): update snapshots
This commit is contained in:
ci[bot]
@@ -6,15 +6,15 @@ General workflow tools, not code-specific.
|
||||
|
||||
Reachable only when you type them (Claude Code: `disable-model-invocation: true`; Codex: `policy.allow_implicit_invocation: false` in `agents/openai.yaml`).
|
||||
|
||||
- **[grill-me](./grill-me/SKILL.md)** — Get relentlessly interviewed about a plan or design until every branch of the design tree is resolved.
|
||||
- **[handoff](./handoff/SKILL.md)** — Compact the current conversation into a handoff document so another agent can continue the work.
|
||||
- **[teach](./teach/SKILL.md)** — Teach the user a new skill or concept over multiple sessions, using the current directory as a stateful teaching workspace.
|
||||
- **[to-questionnaire](./to-questionnaire/SKILL.md)** — Turn a decision you can't answer alone into a Markdown questionnaire for the one person who can — filled in async, or together over a meeting.
|
||||
- **[wait-what](./wait-what/SKILL.md)** — Fire this the moment a message doesn't land. The agent re-pitches it with the context you're missing, in plain English, using your `CONTEXT.md` vocabulary.
|
||||
- **[grill-me](./grill-me/SKILL.md)**: Get relentlessly interviewed about a plan or design until every branch of the design tree is resolved.
|
||||
- **[handoff](./handoff/SKILL.md)**: Compact the current conversation into a handoff document so another agent can continue the work.
|
||||
- **[teach](./teach/SKILL.md)**: Teach the user a new skill or concept over multiple sessions, using the current directory as a stateful teaching workspace.
|
||||
- **[to-questionnaire](./to-questionnaire/SKILL.md)**: Turn a decision you can't answer alone into a Markdown questionnaire for the one person who can (filled in async, or together over a meeting).
|
||||
- **[wait-what](./wait-what/SKILL.md)**: Fire this the moment a message doesn't land. The agent re-pitches it with the context you're missing, in plain English, using your `CONTEXT.md` vocabulary.
|
||||
|
||||
## Model-invoked
|
||||
|
||||
Model- or user-reachable (rich trigger phrasing so the model can reach for them).
|
||||
|
||||
- **[grilling](./grilling/SKILL.md)** — Interview the user relentlessly about a plan, decision, or idea until every branch of the design tree is resolved.
|
||||
- **[writing-for-agents](./writing-for-agents/SKILL.md)** — Writing documents for agents: skills, AGENTS.md/CLAUDE.md, and any doc an agent reaches by a pointer.
|
||||
- **[grilling](./grilling/SKILL.md)**: Interview the user relentlessly about a plan, decision, or idea until every branch of the design tree is resolved.
|
||||
- **[writing-for-agents](./writing-for-agents/SKILL.md)**: Writing documents for agents: skills, AGENTS.md/CLAUDE.md, and any doc an agent reaches by a pointer.
|
||||
|
||||
@@ -4,4 +4,4 @@ description: A relentless interview to sharpen a plan or design.
|
||||
disable-model-invocation: true
|
||||
---
|
||||
|
||||
Run a `/grilling` session.
|
||||
Call the Skill tool with "grilling".
|
||||
|
||||
@@ -5,18 +5,24 @@ description: Grill the user relentlessly about a plan, decision, or idea. Use wh
|
||||
|
||||
Interview the user relentlessly until you reach a shared understanding. Map this as a **design tree**: every decision branches into the decisions that hang off it.
|
||||
|
||||
Work the tree in **rounds**. The **frontier** is every decision whose prerequisites are already settled — the questions you can ask _now_ without guessing at answers you haven't heard yet. Ask the whole frontier in one round: number each question and give your recommended answer. Then wait for the user's answers before the next round.
|
||||
Work the tree in **rounds**. The **frontier** is every decision whose prerequisites are already settled: the questions you can ask _now_ without guessing at answers you haven't heard yet. Ask the whole frontier in one round: number each question and give your recommended answer. Then wait for the user's answers before the next round.
|
||||
|
||||
Each question should be formatted like so:
|
||||
Format a round like so:
|
||||
|
||||
```
|
||||
❓ **Q1** - **<question title>**: <question body, might be multiple paragraphs, including multiple choices>
|
||||
|
||||
➡️ <your recommended answer>
|
||||
|
||||
---
|
||||
|
||||
❓ **Q2** - **<question title>**: <question body, might be multiple paragraphs, including multiple choices>
|
||||
|
||||
➡️ <your recommended answer>
|
||||
```
|
||||
|
||||
Each round the user answers reshapes the tree — settled decisions push the frontier outward and unblock questions that depended on them. Recompute the frontier and ask the next round. A question whose answer depends on another question still open in this round belongs to a _later_ round, not this one.
|
||||
Each round the user answers reshapes the tree: settled decisions push the frontier outward and unblock questions that depended on them. Recompute the frontier and ask the next round. A question whose answer depends on another question still open in this round belongs to a _later_ round, not this one.
|
||||
|
||||
Finding _facts_ is your job, never the user's. When a frontier question needs a fact from the environment (filesystem, tools, etc.), dispatch a sub-agent to find it — don't ask the user for anything you could look up yourself. Don't block on it: a running exploration is an unsettled prerequisite, so only the questions downstream of it wait for the sub-agent to report — ask the rest of the frontier now. The _decisions_ are the user's — put each to them and wait.
|
||||
Finding _facts_ is your job, never the user's. When a frontier question needs a fact from the environment (filesystem, tools, etc.), dispatch a sub-agent to find it; don't ask the user for anything you could look up yourself. Don't block on it: a running exploration is an unsettled prerequisite, so only the questions downstream of it wait for the sub-agent to report; ask the rest of the frontier now. The _decisions_ are the user's: put each to them and wait.
|
||||
|
||||
The session is done when the frontier is empty: every branch of the design tree visited, nothing left silently assumed. Do not act on it until the user confirms you have reached a shared understanding.
|
||||
|
||||
@@ -7,7 +7,7 @@ disable-model-invocation: true
|
||||
|
||||
Write a handoff document summarising the current conversation so a fresh agent can continue the work. Save to the temporary directory of the user's OS - not the current workspace.
|
||||
|
||||
Include a "suggested skills" section in the document, which suggests skills that the agent should invoke.
|
||||
Include a "suggested skills" section in the document, naming which skills the next agent should call the Skill tool for.
|
||||
|
||||
Do not duplicate content already captured in other artifacts (specs, plans, ADRs, issues, commits, diffs). Reference them by path or URL instead.
|
||||
|
||||
|
||||
@@ -16,7 +16,7 @@ Muscle growth driven by mechanical tension and metabolic stress over repeated tr
|
||||
_Avoid_: Bulking, getting big
|
||||
|
||||
**Progressive overload**:
|
||||
Systematically increasing the demand on a muscle over time — via load, volume, or intensity.
|
||||
Systematically increasing the demand on a muscle over time, via load, volume, or intensity.
|
||||
_Avoid_: Pushing harder, levelling up
|
||||
|
||||
**RPE (Rate of Perceived Exertion)**:
|
||||
@@ -29,7 +29,7 @@ _Avoid_: Effort score, intensity rating
|
||||
- **Add a term only when the user understands it.** The glossary is a record of compressed knowledge, not a dictionary the user reads to learn. If the user has just been introduced to a concept, wait until they can use it correctly before promoting it here.
|
||||
- **Be opinionated.** When several words exist for the same concept, pick the best one and list the rest as aliases to avoid. This is how language compresses.
|
||||
- **Keep definitions tight.** One or two sentences. Define what the term IS, not what it does or how to do it.
|
||||
- **Use the glossary's own terms inside definitions.** Once a term is in the glossary, prefer it everywhere — including inside other definitions. This is what makes complex terms easier to grasp later.
|
||||
- **Use the glossary's own terms inside definitions.** Once a term is in the glossary, prefer it everywhere, including inside other definitions. This is what makes complex terms easier to grasp later.
|
||||
- **Group under subheadings** when natural clusters emerge (e.g. `## Anatomy`, `## Programming`). A flat list is fine when terms cohere.
|
||||
- **Flag ambiguities explicitly.** If a term is used loosely in the wider field, note the resolution: "In this workspace, 'set' always means a working set — warm-ups are tracked separately."
|
||||
- **Flag ambiguities explicitly.** If a term is used loosely in the wider field, note the resolution: "In this workspace, 'set' always means a working set; warm-ups are tracked separately."
|
||||
- **Revise as understanding deepens.** A definition the user wrote in week one may be wrong by week six. Update in place; do not leave stale entries.
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# Learning Record Format
|
||||
|
||||
Learning records live in `./learning-records/` and use sequential numbering: `0001-slug.md`, `0002-slug.md`, etc. Create the directory lazily — only when the first record is written.
|
||||
Learning records live in `./learning-records/` and use sequential numbering: `0001-slug.md`, `0002-slug.md`, etc. Create the directory lazily: only when the first record is written.
|
||||
|
||||
They are the teaching equivalent of ADRs: they capture non-obvious lessons, key insights, and stated prior knowledge that will steer future sessions. They are used to calculate the zone of proximal development.
|
||||
|
||||
@@ -12,15 +12,15 @@ They are the teaching equivalent of ADRs: they capture non-obvious lessons, key
|
||||
{1-3 sentences: what was learned (or what prior knowledge was established), and why it matters for future sessions.}
|
||||
```
|
||||
|
||||
That is the whole format. A learning record can be a single paragraph. The value is recording _that_ this is now known and _why_ it changes what to teach next — not in filling out sections.
|
||||
That is the whole format. A learning record can be a single paragraph. The value is recording _that_ this is now known and _why_ it changes what to teach next, not in filling out sections.
|
||||
|
||||
## Optional sections
|
||||
|
||||
Only include these when they add genuine value. Most records won't need them.
|
||||
|
||||
- **Status** frontmatter (`active | superseded by LR-NNNN`) — useful when an earlier understanding turns out to be wrong and is replaced.
|
||||
- **Evidence** — how the user demonstrated the understanding (a question answered, an exercise completed, prior experience cited). Useful when the claim might be revisited.
|
||||
- **Implications** — what this unlocks or rules out for future sessions. Worth recording when non-obvious.
|
||||
- **Status** frontmatter (`active | superseded by LR-NNNN`): useful when an earlier understanding turns out to be wrong and is replaced.
|
||||
- **Evidence**: how the user demonstrated the understanding (a question answered, an exercise completed, prior experience cited). Useful when the claim might be revisited.
|
||||
- **Implications**: what this unlocks or rules out for future sessions. Worth recording when non-obvious.
|
||||
|
||||
## Numbering
|
||||
|
||||
@@ -30,16 +30,16 @@ Scan `./learning-records/` for the highest existing number and increment by one.
|
||||
|
||||
Write one when any of these is true:
|
||||
|
||||
1. **The user demonstrated genuine understanding of something non-trivial** — not just exposure, but evidence they can use the concept correctly. This sets a new floor for what to teach next.
|
||||
2. **The user disclosed prior knowledge** — "I already know X." Record it so future sessions don't re-teach it. Also record the _depth_ claimed.
|
||||
3. **A misconception was corrected** — the user previously believed something wrong and now sees why. These are high-value: they predict future stumbling blocks for related topics.
|
||||
4. **The mission shifted in response to learning** — the user discovered they cared about something different than they thought. Cross-link to [[MISSION.md]] and update it.
|
||||
1. **The user demonstrated genuine understanding of something non-trivial**: not just exposure, but evidence they can use the concept correctly. This sets a new floor for what to teach next.
|
||||
2. **The user disclosed prior knowledge**: "I already know X." Record it so future sessions don't re-teach it. Also record the _depth_ claimed.
|
||||
3. **A misconception was corrected**: the user previously believed something wrong and now sees why. These are high-value: they predict future stumbling blocks for related topics.
|
||||
4. **The mission shifted in response to learning**: the user discovered they cared about something different than they thought. Cross-link to [[MISSION.md]] and update it.
|
||||
|
||||
### What does _not_ qualify
|
||||
|
||||
- Material that was merely covered. Coverage is not learning. Wait for evidence.
|
||||
- Anything already captured tersely in [[GLOSSARY.md]] as a term definition. Don't duplicate.
|
||||
- Session-by-session activity logs. Learning records are not a journal — they are decision-grade insights.
|
||||
- Session-by-session activity logs. Learning records are not a journal: they are decision-grade insights.
|
||||
|
||||
## Supersession
|
||||
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# MISSION.md Format
|
||||
|
||||
`MISSION.md` lives at the workspace root. It captures the _reason_ the user is learning this topic. Every teaching decision — what to teach next, which resources to surface, which exercises to design — should trace back to this document.
|
||||
`MISSION.md` lives at the workspace root. It captures the _reason_ the user is learning this topic. Every teaching decision (what to teach next, which resources to surface, which exercises to design) should trace back to this document.
|
||||
|
||||
## Template
|
||||
|
||||
@@ -8,7 +8,7 @@
|
||||
# Mission: {Topic}
|
||||
|
||||
## Why
|
||||
{1-3 sentences. The concrete real-world goal the user is chasing. What changes in their life or work when they have this skill? Avoid abstract framings like "to understand X" — push for the underlying outcome.}
|
||||
{1-3 sentences. The concrete real-world goal the user is chasing. What changes in their life or work when they have this skill? Avoid abstract framings like "to understand X"; push for the underlying outcome.}
|
||||
|
||||
## Success looks like
|
||||
- {A specific, observable thing the user will be able to do}
|
||||
@@ -19,7 +19,7 @@
|
||||
- {Time, budget, prior commitments, learning preferences, anything that bounds the approach}
|
||||
|
||||
## Out of scope
|
||||
- {Adjacent topics the user explicitly does not want to chase right now — protects the zone of proximal development}
|
||||
- {Adjacent topics the user explicitly does not want to chase right now, protecting the zone of proximal development}
|
||||
```
|
||||
|
||||
## Rules
|
||||
@@ -27,5 +27,5 @@
|
||||
- **One mission per workspace.** If the user wants to learn two unrelated things, that is two workspaces.
|
||||
- **Concrete over abstract.** "Run a half marathon by October" beats "get fitter." "Ship a Rust CLI to my team" beats "learn Rust."
|
||||
- **Push back on vagueness.** If the user cannot articulate why, interview them before writing anything. A bad mission is worse than no mission.
|
||||
- **Revise when reality shifts.** Missions change. When the user's goal moves, update this file — don't leave a stale mission steering future sessions.
|
||||
- **Revise when reality shifts.** Missions change. When the user's goal moves, update this file: don't leave a stale mission steering future sessions.
|
||||
- **Keep it short.** If `MISSION.md` runs past a screen, it has stopped being a compass and started being a plan.
|
||||
|
||||
@@ -9,9 +9,9 @@
|
||||
|
||||
## Knowledge
|
||||
|
||||
- [Book: _The Science and Practice of Strength Training_ — Zatsiorsky & Kraemer](https://example.com)
|
||||
- [Book: _The Science and Practice of Strength Training_ by Zatsiorsky & Kraemer](https://example.com)
|
||||
Foundational text on programming and adaptation. Use for: anything to do with periodisation, recovery, intensity zones.
|
||||
- [Article: "How Much Should I Train?" — Greg Nuckols (Stronger By Science)](https://example.com)
|
||||
- [Article: "How Much Should I Train?" by Greg Nuckols (Stronger By Science)](https://example.com)
|
||||
Evidence-based review of volume landmarks. Use for: weekly set targets per muscle group.
|
||||
|
||||
## Wisdom (Communities)
|
||||
|
||||
@@ -46,9 +46,9 @@ Fluency can give the user an illusory sense of mastery, but storage strength is
|
||||
|
||||
## Lessons
|
||||
|
||||
A lesson is the main thing you produce — the unit in which knowledge and skills reach the user. Each lesson is one self-contained HTML file, saved to `./lessons/` and titled `0001-<dash-case-name>.html` where the number increments each time.
|
||||
A lesson is the main thing you produce: the unit in which knowledge and skills reach the user. Each lesson is one self-contained HTML file, saved to `./lessons/` and titled `0001-<dash-case-name>.html` where the number increments each time.
|
||||
|
||||
A lesson should be **beautiful** — clean, readable typography and layout — since the user will return to these later to review. Think Tufte.
|
||||
A lesson should be **beautiful**, with clean, readable typography and layout, since the user will return to these later to review. Think Tufte.
|
||||
|
||||
The lesson should be short, and completable very quickly. Learners' working memory is very small, and we need to stay within it. But each lesson should give the user a single tangible win that they can build on. It should be directly tied to the mission, and should be in the user's zone of proximal development.
|
||||
|
||||
@@ -62,9 +62,9 @@ Each lesson should contain a reminder to ask followup questions to the agent. Th
|
||||
|
||||
## Assets
|
||||
|
||||
Lessons are built from reusable **components**, stored in `./assets/`: stylesheets, quiz widgets, simulators, diagram helpers — anything a second lesson could reuse.
|
||||
Lessons are built from reusable **components**, stored in `./assets/`: stylesheets, quiz widgets, simulators, diagram helpers, and anything else a second lesson could reuse.
|
||||
|
||||
Reuse is the default, not the exception. Before authoring a lesson, read `./assets/` and build from the components already there. When a lesson needs something new and reusable, write it as a component in `./assets/` and link to it — never inline code a future lesson would duplicate.
|
||||
Reuse is the default, not the exception. Before authoring a lesson, read `./assets/` and build from the components already there. When a lesson needs something new and reusable, write it as a component in `./assets/` and link to it; never inline code a future lesson would duplicate.
|
||||
|
||||
A shared stylesheet is the first component every workspace earns: every lesson links it, so the lessons look like one consistent course rather than a pile of one-offs. As the workspace grows, so should the component library.
|
||||
|
||||
|
||||
@@ -4,10 +4,11 @@ description: Turn a decision you can't fully answer into a questionnaire for som
|
||||
disable-model-invocation: true
|
||||
---
|
||||
|
||||
Turn something the user can't answer alone into a **questionnaire** — a Markdown document they hand to one person to fill in async, or fill out together over a meeting. The recipient holds knowledge the user lacks; the questionnaire pulls it out of them.
|
||||
Turn something the user can't answer alone into a **questionnaire**: a Markdown document they hand to one person to fill in async, or fill out together over a meeting. The recipient holds knowledge the user lacks; the questionnaire pulls it out of them.
|
||||
|
||||
**Grill the send, not the subject.** Interview the user only about the _send_, which they can always answer: who it goes to, and what they need back. The questions in the document then target the **gap** between what the recipient knows and what the user needs.
|
||||
|
||||
|
||||
1. **Who is it going to?** Ask, in one exchange, the recipient's role, expertise, and relationship to the user. This fixes the questionnaire's tone and how much context it must carry. Done when you know who the recipient is and what they know that the user doesn't.
|
||||
|
||||
2. **What do you need back?** Ask, in one exchange, the specific decisions or facts the user can't resolve alone and needs from this person. Done when you have a concrete list of what the user must walk away able to do or decide.
|
||||
@@ -16,7 +17,7 @@ Turn something the user can't answer alone into a **questionnaire** — a Markdo
|
||||
|
||||
## Document structure
|
||||
|
||||
Frame the document as a **discovery questionnaire**: the user lacks context, the recipient holds it. Order questions most-important-first — async means you may only get one pass — and group them under `##` headings by theme once there are more than a handful. Write it using the template below.
|
||||
Frame the document as a **discovery questionnaire**: the user lacks context, the recipient holds it. Order questions most-important-first, since async means you may only get one pass, and group them under `##` headings by theme once there are more than a handful. Write it using the template below.
|
||||
|
||||
<questionnaire-template>
|
||||
|
||||
@@ -24,7 +25,7 @@ Frame the document as a **discovery questionnaire**: the user lacks context, the
|
||||
|
||||
**Purpose:** why this questionnaire exists and the decision riding on it.
|
||||
|
||||
**From:** <the user> — **To:** <the recipient> — **How your answers will be used:** <where they go>
|
||||
**From:** <the user>, **To:** <the recipient>, **How your answers will be used:** <where they go>
|
||||
|
||||
## Context
|
||||
|
||||
@@ -32,11 +33,11 @@ One paragraph orienting a recipient who wasn't in the user's head. Enough to ans
|
||||
|
||||
## How to answer
|
||||
|
||||
Deadline and rough effort. Partial answers and "I don't know" are useful — flag anything you're unsure of rather than skipping it.
|
||||
Deadline and rough effort. Partial answers and "I don't know" are useful: flag anything you're unsure of rather than skipping it.
|
||||
|
||||
## <Theme heading>
|
||||
|
||||
One `##` section per theme. Under each, its questions, most-important-first. Every question is one idea — never compound — with an answer stub directly beneath, and a one-line _why this matters_ only where the question could be misread or invite a throwaway answer.
|
||||
One `##` section per theme. Under each, its questions, most-important-first. Every question is one idea, never compound, with an answer stub directly beneath, and a one-line _why this matters_ only where the question could be misread or invite a throwaway answer.
|
||||
|
||||
<question-example>
|
||||
### What load is the system expected to handle at launch?
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
---
|
||||
name: wait-what
|
||||
description: Stop. That last message did not land — re-pitch it.
|
||||
description: "Stop. That last message did not land: re-pitch it."
|
||||
disable-model-invocation: true
|
||||
---
|
||||
|
||||
Wait — I don't understand where you've got to here. Re-pitch that: give me a little bit of context, talk in ASD-STE100 Simplified Technical English, and use the ubiquitous language from `CONTEXT.md`.
|
||||
Wait, I don't understand where you've got to here. Re-pitch that: give me a little bit of context, talk in ASD-STE100 Simplified Technical English, and use the ubiquitous language from `CONTEXT.md` (follow `CONTEXT-MAP.md` to the right one if the repo has more than one).
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
interface:
|
||||
display_name: "Wait What"
|
||||
short_description: "Re-pitch that — simpler, with the context I'm missing"
|
||||
short_description: "Re-pitch that: simpler, with the context I'm missing"
|
||||
policy:
|
||||
allow_implicit_invocation: false
|
||||
|
||||
@@ -1,21 +1,21 @@
|
||||
# Skill mechanics
|
||||
|
||||
The skill-specific branch of [`writing-for-agents`](SKILL.md): what changes when the document is a skill — frontmatter, the invocation choice, and router skills. Everything else about writing it is the universal reference in `SKILL.md`.
|
||||
The skill-specific branch of [`writing-for-agents`](SKILL.md): what changes when the document is a skill (frontmatter, the invocation choice, and router skills). Everything else about writing it is the universal reference in `SKILL.md`.
|
||||
|
||||
## Invocation
|
||||
|
||||
Two choices, trading the two loads:
|
||||
|
||||
- A **model-invoked** skill keeps a `description`, so the agent can fire it autonomously — and other skills can reach it. You can still type its name: model-invocation always _includes_ user reach; a description only ever adds agent discovery, never removes the human's. The description is the skill's top-level context pointer, forced to stay loaded at all times — permanent context load in exchange for discoverability. A model-invoked skill whose content is all reference is also one home for shared reference: another skill can invoke it, so reference needed by several skills lives in one place. Mechanics: omit `disable-model-invocation`, and write a model-facing description carrying the trigger branches (the pointer-writing rules in `SKILL.md` apply in full).
|
||||
- A **user-invoked** skill strips the description from the agent's reach: only the human typing its name can invoke it, and no other skill can. Zero context load, but it spends cognitive load — you are the index that must remember it exists. Mechanics: set `disable-model-invocation: true`; the `description` becomes human-facing — a one-line summary, trigger lists stripped.
|
||||
- A **model-invoked** skill keeps a `description`, so the agent can fire it autonomously, and other skills can reach it. You can still type its name: model-invocation always _includes_ user reach; a description only ever adds agent discovery, never removes the human's. The description is the skill's top-level context pointer, forced to stay loaded at all times: permanent context load in exchange for discoverability. A model-invoked skill whose content is all reference is also one home for shared reference: another skill can invoke it, so reference needed by several skills lives in one place. Mechanics: omit `disable-model-invocation`, and write a model-facing description carrying the trigger branches (the pointer-writing rules in `SKILL.md` apply in full).
|
||||
- A **user-invoked** skill strips the description from the agent's reach: only the human typing its name can invoke it, and no other skill can. Zero context load, but it spends cognitive load: you are the index that must remember it exists. Mechanics: set `disable-model-invocation: true`; the `description` becomes human-facing: a one-line summary, trigger lists stripped.
|
||||
|
||||
Pick model-invocation only when the agent must reach the skill on its own, or another skill must. If it only ever fires by hand, make it user-invoked and pay no context load.
|
||||
|
||||
Shared reference that two user-invoked skills both need can live in neither — with no descriptions, neither can fire the other. Push it to a plain file outside the skill system: external reference any skill can point at.
|
||||
Shared reference that two user-invoked skills both need can live in neither: with no descriptions, neither can fire the other. Push it to a plain file outside the skill system: external reference any skill can point at.
|
||||
|
||||
## Splitting by invocation
|
||||
|
||||
The invocation cut of splitting (the sequence cut lives in `SKILL.md`): split off a model-invoked skill when you have a distinct leading word that should trigger it on its own — a trigger word you actually use in your prompts — or another skill must reach it. You pay context load for the new always-loaded description, so that independent reach has to be worth it.
|
||||
The invocation cut of splitting (the sequence cut lives in `SKILL.md`): split off a model-invoked skill when you have a distinct leading word that should trigger it on its own (a trigger word you actually use in your prompts), or another skill must reach it. You pay context load for the new always-loaded description, so that independent reach has to be worth it.
|
||||
|
||||
## Router skills
|
||||
|
||||
|
||||
@@ -3,17 +3,17 @@ name: writing-for-agents
|
||||
description: Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.
|
||||
---
|
||||
|
||||
Reference for writing any document an agent consumes — a skill, an `AGENTS.md` / `CLAUDE.md`, a doc reached by a pointer. The packaging differs; the writing does not: the same levers make each one predictable — the agent taking the same _process_ every run, not producing the same output.
|
||||
Reference for writing any document an agent consumes: a skill, an `AGENTS.md` / `CLAUDE.md`, a doc reached by a pointer. The packaging differs; the writing does not: the same levers make each one predictable, since the agent takes the same _process_ every run rather than producing the same output.
|
||||
|
||||
When the document you're writing is a skill, read [`SKILL-MECHANICS.md`](SKILL-MECHANICS.md) for frontmatter, invocation choice, and router skills.
|
||||
|
||||
## Context pointers
|
||||
|
||||
A **context pointer** is a reference held in the agent's context that names some out-of-context material and encodes the condition for reaching it. A skill's description is one; a line in `AGENTS.md` naming a doc is the same object. The pointer's _wording_, not its target, decides when the agent reaches the material — and how reliably. A must-have target behind a weakly worded pointer is a variance bug: sharpen the wording first, and inline the material only if sharpening fails.
|
||||
A **context pointer** is a reference held in the agent's context that names some out-of-context material and encodes the condition for reaching it. A skill's description is one; a line in `AGENTS.md` naming a doc is the same object. The pointer's _wording_, not its target, decides when the agent reaches the material, and how reliably. A must-have target behind a weakly worded pointer is a variance bug: sharpen the wording first, and inline the material only if sharpening fails.
|
||||
|
||||
A pointer does two jobs — state what the material is, and list the **branches** that should trigger reaching it (a branch is a distinct case the document handles, so different runs take different paths through it). Every word of an always-loaded pointer costs on every turn, so it earns even harder pruning than the body:
|
||||
A pointer does two jobs: state what the material is, and list the **branches** that should trigger reaching it (a branch is a distinct case the document handles, so different runs take different paths through it). Every word of an always-loaded pointer costs on every turn, so it earns even harder pruning than the body:
|
||||
|
||||
- **Front-load the leading word** — the pointer is where it does its triggering work.
|
||||
- **Front-load the leading word**: the pointer is where it does its triggering work.
|
||||
- **One trigger per branch.** Synonyms that rename a single branch are one branch written twice; collapse them and keep only genuinely distinct branches.
|
||||
- **Cut identity the body already carries.**
|
||||
|
||||
@@ -21,33 +21,33 @@ A pointer does two jobs — state what the material is, and list the **branches*
|
||||
|
||||
Every document and pointer you add spends one of two budgets:
|
||||
|
||||
- **Context load** — the cost of always-loaded material on the agent's window: an `AGENTS.md` line, a skill description, anything sitting in context every turn, spending tokens and attention whether or not it fires.
|
||||
- **Cognitive load** — the cost on the human: which documents exist and when to reach for each. The human is the index. Not a cost to minimise — it is the price of human agency; spend it where human judgement matters, remove it where it does not.
|
||||
- **Context load** is the cost of always-loaded material on the agent's window: an `AGENTS.md` line, a skill description, anything sitting in context every turn, spending tokens and attention whether or not it fires.
|
||||
- **Cognitive load** is the cost on the human: which documents exist and when to reach for each. The human is the index. Not a cost to minimise: it is the price of human agency; spend it where human judgement matters, remove it where it does not.
|
||||
|
||||
Material reached only through a pointer escapes context load at the price of the pointer's own line; material with no pointer at all rides entirely on cognitive load.
|
||||
|
||||
## Information hierarchy
|
||||
|
||||
A document is built from two content types — **steps** (the ordered actions the agent performs) and **reference** (definitions, rules, facts consulted on demand) — that mix freely: all steps (a recipe), all reference (a review's rules, this skill), or both. The core decision is where each piece sits on the **information hierarchy**, a ladder ranked by how immediately the agent needs the material:
|
||||
A document is built from two content types: **steps** (the ordered actions the agent performs) and **reference** (definitions, rules, facts consulted on demand). The two mix freely: all steps (a recipe), all reference (a review's rules, this skill), or both. The core decision is where each piece sits on the **information hierarchy**, a ladder ranked by how immediately the agent needs the material:
|
||||
|
||||
1. **In-file step** — the primary tier: what the agent does, in order.
|
||||
2. **In-file reference** — consulted on demand. Often a legitimately flat peer-set (every rule of a review on one rung) — a fine arrangement, not a smell.
|
||||
3. **Disclosed reference** — pushed out into a separate file, reached by a context pointer, loaded only when the pointer fires. Spans a sibling file in the same folder through fully external reference that lives anywhere and any document can point at.
|
||||
1. **In-file step** is the primary tier: what the agent does, in order.
|
||||
2. **In-file reference** is consulted on demand. Often a legitimately flat peer-set (every rule of a review on one rung), which is a fine arrangement, not a smell.
|
||||
3. **Disclosed reference** is pushed out into a separate file, reached by a context pointer, loaded only when the pointer fires. Spans a sibling file in the same folder through fully external reference that lives anywhere and any document can point at.
|
||||
|
||||
Push too little down and the top bloats; push too much and you hide material the agent actually needs. That tension is the whole decision.
|
||||
|
||||
**Progressive disclosure** is the move down the ladder — out of the main file and behind a pointer — so the top stays legible. Not primarily a token optimisation: it is how the hierarchy is protected. Branching is the cleanest disclosure test: inline what every branch needs, and push behind a pointer what only some branches reach. When a document has steps, in-file reference that should be disclosed buries them and turns attending to them into a coin-flip — a variance lever, not just a legibility one.
|
||||
**Progressive disclosure** is the move down the ladder (out of the main file and behind a pointer) so the top stays legible. Not primarily a token optimisation: it is how the hierarchy is protected. Branching is the cleanest disclosure test: inline what every branch needs, and push behind a pointer what only some branches reach. When a document has steps, in-file reference that should be disclosed buries them and turns attending to them into a coin-flip: a variance lever, not just a legibility one.
|
||||
|
||||
**Co-location** is the within-file companion: where the ladder decides _how far down_ a piece sits, co-location decides _what sits beside it_ once there. Keep a concept's definition, rules, and caveats under one heading rather than scattered, so reading one part brings its neighbours with it. The test: the document should read like documentation written for the agent — grouped material reads that way; scattered material does not. (Distinct from duplication: that repeats one meaning in two places; scattering fragments one meaning across many.)
|
||||
**Co-location** is the within-file companion: where the ladder decides _how far down_ a piece sits, co-location decides _what sits beside it_ once there. Keep a concept's definition, rules, and caveats under one heading rather than scattered, so reading one part brings its neighbours with it. The test: the document should read like documentation written for the agent. Grouped material reads that way; scattered material does not. (Distinct from duplication: that repeats one meaning in two places; scattering fragments one meaning across many.)
|
||||
|
||||
**Sprawl** is the failure mode here: a document simply too long, even when every line is live and unique. Attention thins across the excess, and every extra line is one more to keep relevant. The cure is the ladder: disclose reference behind pointers, and split by branch or sequence so each path carries only what it needs.
|
||||
|
||||
## Steps and completion criteria
|
||||
|
||||
Every step ends on a **completion criterion** — the condition that tells the agent the work is done. Two properties make it a lever:
|
||||
Every step ends on a **completion criterion**, the condition that tells the agent the work is done. Two properties make it a lever:
|
||||
|
||||
- **Clarity** — can the agent tell done from not-done? A vague bound ("understanding reached") invites **premature completion**: ending the step before it is genuinely done, attention slipping to _being done_. The visible steps still ahead — the **post-completion steps** — supply the pull; the criterion's clarity is the resistance. Defend in order: **sharpen the bound first** (local and cheap); only if it is irreducibly fuzzy _and_ you observe the rush, hide the later steps by splitting the sequence — and hiding only works across a real context boundary (a hand-off or a subagent dispatch; an inline call leaves the later steps in context and clears nothing).
|
||||
- **Demand** — how much it requires. "Every modified model accounted for" forces thorough work where "produce a change list" does not. Demand drives **legwork** — the digging the agent does within the work, latent in the wording rather than written as its own step — and it is not step-bound: "every rule applied" binds a body of flat reference just as "every step done" binds a sequence, which is how an all-reference document still carries an exhaustiveness bar.
|
||||
- **Clarity**: can the agent tell done from not-done? A vague bound ("understanding reached") invites **premature completion**: ending the step before it is genuinely done, attention slipping to _being done_. The visible steps still ahead (the **post-completion steps**) supply the pull; the criterion's clarity is the resistance. Defend in order: **sharpen the bound first** (local and cheap); only if it is irreducibly fuzzy _and_ you observe the rush, hide the later steps by splitting the sequence. Hiding only works across a real context boundary (a hand-off or a subagent dispatch; an inline call leaves the later steps in context and clears nothing).
|
||||
- **Demand**: how much it requires. "Every modified model accounted for" forces thorough work where "produce a change list" does not. Demand drives **legwork** (the digging the agent does within the work, latent in the wording rather than written as its own step), and it is not step-bound: "every rule applied" binds a body of flat reference just as "every step done" binds a sequence, which is how an all-reference document still carries an exhaustiveness bar.
|
||||
|
||||
The strongest criteria are both checkable and exhaustive.
|
||||
|
||||
@@ -55,27 +55,27 @@ The strongest criteria are both checkable and exhaustive.
|
||||
|
||||
Splitting one document into two spends one of the two loads, so split only when the cut earns it:
|
||||
|
||||
- **By sequence** — split a run of steps where the post-completion steps tempt the agent to rush the one in front of it. Keeping them out of view drives more legwork on the current task. Beware the reverse: merging sequences exposes each step's later steps to what follows, inviting premature completion.
|
||||
- **By invocation** — skill-specific: see [`SKILL-MECHANICS.md`](SKILL-MECHANICS.md).
|
||||
- **By sequence**: split a run of steps where the post-completion steps tempt the agent to rush the one in front of it. Keeping them out of view drives more legwork on the current task. Beware the reverse: merging sequences exposes each step's later steps to what follows, inviting premature completion.
|
||||
- **By invocation**, skill-specific: see [`SKILL-MECHANICS.md`](SKILL-MECHANICS.md).
|
||||
|
||||
## Leading words
|
||||
|
||||
A **leading word** is a compact concept already living in the model's pretraining that the agent thinks with while running the document (_lesson_, _fog of war_, _tracer bullets_). Repeated as a token, never as a sentence, it accumulates a distributed definition and anchors a whole region of behaviour in the fewest tokens, by recruiting priors the model already holds. Coining your own works if you define it clearly, but a made-up word recruits no priors — you pay in definition tokens what a pretrained word gives free; reach for an existing word first.
|
||||
A **leading word** is a compact concept already living in the model's pretraining that the agent thinks with while running the document (_lesson_, _fog of war_, _tracer bullets_). Repeated as a token, never as a sentence, it accumulates a distributed definition and anchors a whole region of behaviour in the fewest tokens, by recruiting priors the model already holds. Coining your own works if you define it clearly, but a made-up word recruits no priors: you pay in definition tokens what a pretrained word gives free; reach for an existing word first.
|
||||
|
||||
It anchors twice. In the body, _execution_: the agent reaches for the same behaviour every time the word appears, and inside flat reference it focuses attention on a class of thing to look for. In a pointer, _invocation_: when the same word lives in your prompts, your docs, and your codebase, the agent links that shared language to the material and reaches it more reliably.
|
||||
|
||||
Hunt for opportunities to refactor with leading words. A triad spelled out at three sites, a pointer spending a sentence to gesture at one idea — each is a passage begging to collapse into a single token:
|
||||
Hunt for opportunities to refactor with leading words. A triad spelled out at three sites, a pointer spending a sentence to gesture at one idea. Each is a passage begging to collapse into a single token:
|
||||
|
||||
- "fast, deterministic, low-overhead" → _tight_ (a _tight_ loop).
|
||||
- "a loop you believe in" → _red_ — a fuzzy gate becomes a binary observable state (the loop goes _red_ on the bug, or it doesn't).
|
||||
- "a loop you believe in" → _red_, turning a fuzzy gate into a binary observable state (the loop goes _red_ on the bug, or it doesn't).
|
||||
|
||||
You win twice: fewer tokens, and a sharper hook for the agent to hang its thinking on. Assume every document is carrying restatements that leading words retire — go find them.
|
||||
You win twice: fewer tokens, and a sharper hook for the agent to hang its thinking on. Assume every document is carrying restatements that leading words retire. Go find them.
|
||||
|
||||
**Negation** is the failure mode beside this lever: steering by prohibition drags the forbidden behaviour into context and makes it _more_ available, not less. _Don't think of an elephant_, and the elephant is all there is; the negation is a weak modifier the strongly-activated concept overruns, so the ban half-reads as an instruction to do the thing. Prompt the **positive** — state the target behaviour ("write one-line comments") so the banned one is never spoken. A prohibition earns its place only as a hard guardrail you cannot phrase positively; even then, pair it with the positive target so attention lands on what to do.
|
||||
**Negation** is the failure mode beside this lever: steering by prohibition drags the forbidden behaviour into context and makes it _more_ available, not less. _Don't think of an elephant_, and the elephant is all there is; the negation is a weak modifier the strongly-activated concept overruns, so the ban half-reads as an instruction to do the thing. Prompt the **positive**: state the target behaviour ("write one-line comments") so the banned one is never spoken. A prohibition earns its place only as a hard guardrail you cannot phrase positively; even then, pair it with the positive target so attention lands on what to do.
|
||||
|
||||
## Pruning
|
||||
|
||||
- Keep each meaning in a **single source of truth**: one authoritative place, so changing the behaviour is a one-place edit. **Duplication** — the same meaning in more than one place — costs maintenance and tokens, and inflates a meaning's prominence on the ladder past its real rank. (The accidental inverse of a leading word, which repeats a token on purpose, never the meaning.)
|
||||
- The **environment** is a source of truth too — `package.json` scripts, config files, the directory layout, `--help` output — and a document that restates it is a **cache**: a copy of a lookup, earning its load only when the lookup is expensive. Cache what the agent cannot find by looking: the unwritten convention, the reason behind a choice, the gotcha no config confesses. Leave the one-file, one-command lookups to the environment, where they cannot go stale.
|
||||
- Keep each meaning in a **single source of truth**: one authoritative place, so changing the behaviour is a one-place edit. **Duplication** (the same meaning in more than one place) costs maintenance and tokens, and inflates a meaning's prominence on the ladder past its real rank. (The accidental inverse of a leading word, which repeats a token on purpose, never the meaning.)
|
||||
- The **environment** is a source of truth too (`package.json` scripts, config files, the directory layout, `--help` output), and a document that restates it is a **cache**: a copy of a lookup, earning its load only when the lookup is expensive. Cache what the agent cannot find by looking: the unwritten convention, the reason behind a choice, the gotcha no config confesses. Leave the one-file, one-command lookups to the environment, where they cannot go stale.
|
||||
- Check every line for **relevance**: does it still bear on what the document does? A line loses relevance by never bearing on the task (mere exposition, or a branch that should be disclosed) or by going stale as the behaviour or world it describes changes. Shorter documents are easier to keep relevant. Without a pruning discipline the default fate is **sediment**: stale layers that settle because adding feels safe and removing feels risky, until you must core down through them to find what is still live.
|
||||
- Hunt **no-ops** sentence by sentence: an instruction the model already obeys by default pays load to say nothing. The test — does it change behaviour versus the default? — is model-relative, not reader-relative: two people disagreeing about a no-op disagree about the default, and settle it by running the document, not by debate. When a sentence fails, delete the whole sentence rather than trim words from it. The test also grades leading words: a word too weak to beat the default (_be thorough_ when the agent is already thorough-ish) is a no-op, and the fix is a stronger word (_relentless_), not a different technique.
|
||||
- Hunt **no-ops** sentence by sentence: an instruction the model already obeys by default pays load to say nothing. The test (does it change behaviour versus the default?) is model-relative, not reader-relative: two people disagreeing about a no-op disagree about the default, and settle it by running the document, not by debate. When a sentence fails, delete the whole sentence rather than trim words from it. The test also grades leading words: a word too weak to beat the default (_be thorough_ when the agent is already thorough-ish) is a no-op, and the fix is a stronger word (_relentless_), not a different technique.
|
||||
|
||||
Reference in New Issue
Block a user