CD post-training-strategy
为 Agent 模型做后训练或选路线时使用——判断该先补 Mid-training、用 SFT 立协议还是直接上 RL；构造 Mid-training 语料与 SFT 数据、搭建 RL 环境、在 GRPO/PPO/DPO 与蒸馏之间取舍；排查 pass@k 近零、格式不稳就训 RL、组内无奖励差异、训练-推理数值失配、过度训练泛化下降。触发词：后训练、Mid-training、继续预训练、SFT、RL、GRPO、PPO、DPO、拒绝采样、RFT、蒸馏、on-policy、RLVR、LoRA。
为 Agent 模型做后训练或选路线时使用——判断该先补 Mid-training、用 SFT 立协议还是直接上 RL；构造 Mid-training 语料与 SFT 数据、搭建 RL 环境、在 GRPO/PPO/DPO 与蒸馏之间取舍；排查 pass@k 近零、格式不稳就训…
As a process D 48/100 · Unfinished process — weak spots: result and completion, when it triggers, inputs and preconditions
How to improve
- Say in the description WHEN to use the skill ("use when…", example requests): that is the agent's main cue.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 1. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
description-no-whendescription does not say WHEN to use the skill (no "use when")
Process rating: all ten parameters 48/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Failures and branches. Linear process with no failure handling
- 20When it triggers. No condition that starts the skill
- 60Tools and files. Uses tools (web) that frontmatter does not declare
- 100Steps. 50 steps
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 1697 tokens
- 100Running it twice. No mutating operations
- 100Progress reporting. Reports progress
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Output format is not stated: the model decides each time
- +4No input/output examples
- +1No license
- +2Single-language instructions
- +3Description length 248: enough signal without eating the budget
- +4Structure: 13 headings
- +3Step-by-step instructions: 50 items
Quality base 70; lint remarks subtract, signals add up to 100. Result: 68.