Community生產力與協作github.com

growing-assistant

A skill of skills: bootstrap a personal, growing AI assistant in any file-capable Agent. It interviews the user, creates a local assistant workspace (profile, rules, memory, docs, workflows, evals), runs one real workflow end-to-end, then turns corrections into rules and eval cases so the next run is better. Trigger: build my own assistant, personal AI, growing assistant, 会成长的数字助手, 个人助手, 助理工作流.

growing-assistant 是什麼?

growing-assistant is a Claude Code agent skill that a skill of skills: bootstrap a personal, growing AI assistant in any file-capable Agent. It interviews the user, creates a local assistant workspace (profile, rules, memory, docs, workflows, evals), runs one real workflow end-to-end, then turns corrections into rules and eval cases so the next run is better. Trigger: build my own assistant, personal AI, growing assistant, 会成长的数字助手, 个人助手, 助理工作流.

相容平台✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/zhoujinq/growing-assistant-skill/tree/HEAD/skills/growing-assistant

Installed? Explore more 生產力與協作 skills: steipete/gemini, steipete/gh-issues, steipete/skill-creator · View all 6 →

在你喜歡的 AI 中提問

開啟一個已預先載入此 Agent Skill 的新對話。

說明文件

Growing Assistant / 成长助手

这不是一个“帮你做某件事”的技能,是一个“帮你长出一个助手”的技能。 它先帮你搭好底座,再陪你跑通第一个工作流,最后把你的纠正变成规则和评测。

What this is

A meta-skill: it does not replace your resume/email/weekly-review skill. It builds the workspace, memory, rules, and feedback loop that make those skills yours and better over time.

It is deliberately small:

  • one local workspace,
  • one first workflow,
  • one correction loop,
  • one effort switch.

No database server. No platform. No multi-agent ceremony.

Hard rules

  1. One workflow at a time. Do not scaffold five workflows before one works end-to-end.
  2. File-first. Chat is not memory. Anything that must survive a new session goes into a file.
  3. Every fact cites a file. Use path#heading or path:line. No source, no claim.
  4. Unknown is allowed. Write not in docs or 待补; never fill gaps with plausible guesses.
  5. Corrections become rules. “I’ll remember” is not a mechanism. Write a rule + an eval case.
  6. Privacy is a hard gate. Secrets, customer names, and company-internal data go in private/; public output must not contain them.
  7. Stop at gates. G0–G3 exist to prevent a confident half-product.
  8. Build one loop, not a platform. If a feature does not make the first loop more verifiable, cut it.

Modes

ModeWhenOutput
bootstrapNew user / empty workspaceassistant/ scaffold + PROFILE.md + first WORKFLOW.md
runUser gives a real taskoutput/<date>-<task>/report.md + evidence.md + handoff.md
growUser corrects a resultnew rule in RULES.md + new case in evals/ + re-run
exportWorkflow is stablestandalone skills/<name>/SKILL.md + a static case page

If the user asks “build my assistant”, start at bootstrap. If they give a task and a workspace exists, go to run. If they say “that’s wrong / next time do X”, go to grow.

The 5-step loop

1. Profile (5–8 questions)

Ask one at a time, with a recommended default. Stop when you can fill:

  • who the user is and what they do,
  • 3 current goals,
  • one repeated task that wastes time,
  • what “done” looks like for that task,
  • what data/tools are allowed,
  • what must never be shared,
  • preferred language/tone,
  • effort preference (fast vs thorough).

Write PROFILE.md. Do not ask for what is already in the repo.

2. Workspace

Create the file contract below. Use the script:

python scripts/init_assistant.py --root ./assistant --name "My Assistant"

The workspace is the state. If it is not on disk, it does not exist.

3. First workflow

Pick one task that is frequent, painful, and verifiable. Write WORKFLOW.md:

  • trigger,
  • inputs,
  • steps,
  • output contract,
  • acceptance checks,
  • stop conditions.

Run it once. Produce a real artifact, not a chat answer. If it cannot produce an artifact, pick another workflow.

4. Correction

Ask: “What exactly is wrong — fact, spec, or preference?”

  • Fact → fix the source file / citation.
  • Spec → change WORKFLOW.md.
  • Preference → write a rule in RULES.md.

Then add one eval case to evals/. Re-run the same task. The second run must show the correction taking effect.

5. Effort

EffortBehaviourUse when
lowanswer directly; one passsimple, reversible
mediumdraft → self-check → revise oncenormal work
highsolve → check → revise; keep best version; record evidencehigh stakes / long tasks

Record the effort level in the output. High effort without extra checks is just slower low effort.

File contract

assistant/
├─ PROFILE.md          # who, goals, constraints, preferences
├─ RULES.md            # durable rules, one per correction
├─ WORKFLOW.md         # the first workflow
├─ memory/             # dated notes, decisions, open loops
├─ docs/               # source material (read-only)
├─ workflows/          # additional workflows, one file each
├─ evals/              # good/bad cases, one file each
├─ output/             # dated task outputs
├─ skills/             # distilled standalone skills
└─ private/            # secrets/customer data; gitignored

Gates

GatePass condition
G0 ProfilePROFILE.md has who/goals/constraints/privacy/preferences
G1 WorkflowWORKFLOW.md has inputs/output/steps/acceptance/stop
G2 First runartifact + evidence + no uncited factual claim
G3 Growthone new rule + one new eval case; second run passes

Do not skip a gate because the user is in a hurry. A skipped gate is the most expensive shortcut.

References

  • references/01-bootstrap.md — questions + profile template
  • references/02-workspace.md — file contract and naming
  • references/03-run.md — run contract and acceptance
  • references/04-grow.md — correction → rule → eval
  • references/05-export.md — distill a workflow into a standalone skill + case page

Definition of done (v0.1)

A new user can:

  1. install the skill,
  2. answer 5–8 questions,
  3. get a workspace with profile/rules/workflow,
  4. run one real task with a cited artifact,
  5. correct one thing,
  6. see the correction persist in a new session.

If any step needs the skill author to explain it, the skill is not done.

相關技能