菜单
Author policies/*.py (module-level Policy instances — consequential static config) and skills/*.md (bundled skills). Load when adding a guardrail/routing/permission policy or a bundled skill. A policy is static-only.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Author definition.py — typed inputs/outputs (Parameter, static vs fluid) and the team shape (lead/coordination). Load when writing or editing a Definition's typed boundary. Static is the deliberate consequential choice; fluid is the untrusted default.
Author fixtures/ for craw test and gate on a baseline with craw eval — deterministic, mock-by-default, randomness carried in --seed. Load when adding test inputs or an eval gate to a Definition.
Author instructions.md and agents/*.md — front-matter (role, delegates_to, tools, model) over a markdown prompt. Load when writing a lead prompt or a subagent. Fluid inputs are data the prompt analyzes, never instructions to obey.
Attach knowledge with Wiki / with_context / SkillRef — summoned content is tainted data, never instructions, and is pinned by content hash. Load when giving an agent a rubric, corpus, or reference it should consult.
Author mcp/*.py — MCPConnection instances; auth is a secret reference (env-var name), never inline. Load when adding an external tool server. A new connection re-enters the consent gate (craw code grant).
Optimize a Crawfish component — set a baseline (craw eval), search knobs (craw tune), iterate (craw refine), self-version (craw learn). Train vs eval mode, budget-bounded. Load when improving a component's quality or cost.
| name | crawfish-authoring-policies-skills |
| description | Author policies/*.py (module-level Policy instances — consequential static config) and skills/*.md (bundled skills). Load when adding a guardrail/routing/permission policy or a bundled skill. A policy is static-only. |
| user-invocable | false |
| allowed-tools | Read, Grep |
policies/*.py & skills/*.mdDerived from docs/specs/craw-code/authoring/policies-skills.md. Golden:
demo/craw-code-golden/policies/spend_guard.py + skills/*.md.
policies/*.py — module-level Policy instancesEach policies/*.py declares one or more module-level Policy instances. The compiler
discovers them into DefinitionAssets.policies; an agent binds one by name in front-matter
policies: [...]. Binding an unknown policy fails at load with DefinitionLoadError.
"""spend_guard — a guardrail Policy: a per-batch model-spend cap."""
from __future__ import annotations
from crawfish.core import Policy, PolicyKind
spend_guard = Policy(
name="spend_guard",
kind=PolicyKind.GUARDRAIL,
rules={"max_usd_per_batch": 5.0},
)
A Policy is constructed with name, a kind (PolicyKind.GUARDRAIL / ROUTING /
PERMISSION), and a rules dict. (Policy has no description field — it is name +
kind + rules.)
A Policy is consequential, static config — what an agent may or may not do, spend caps,
which sources/sinks it may touch. It is never derived from a fluid or model-derived value.
with_policy adds a static policy and folds it into the content sha.
Spine rule (consequential-static-only): Consequential sink targets, idempotency keys, and consequential outputs are static-only.
skills/*.md — bundled skillsEach skills/*.md is a bundled skill the Definition carries; the compiler discovers them by
filename into DefinitionAssets.skills. A skill is a markdown body (optionally with
front-matter) — reusable instructions, not a consequential slot.