聘才猫（Pincaimao）平台基础能力

聘才猫平台基础能力 Use when calling any Pincaimao platform API — file upload, presigned URL, conversation list, message history, audio-to-text, resume JSON upload, or...

Security Scan

The skill's instructions, required binaries, and single API key requirement are consistent with a platform-integration helper for Pincaimao; nothing requested appears out of scope or unexplained.

Purpose & Capability

Name/description match the documented endpoints (chat, file upload, presigned URL, transcripts, resume JSON, conversations). Declared requirements (curl, python3, PCM_ANY_KEY) are appropriate for the documented HTTP calls and streaming-parsing examples.

Instruction Scope

SKILL.md contains concrete curl examples and a small python streaming parser that only processes curl's SSE output. It does not instruct reading unrelated system files, harvesting other env vars, or posting data to unrelated external endpoints. Examples reference file paths only as upload inputs.

Install Mechanism

Instruction-only skill with no install spec and no archive downloads; nothing is written to disk by an installer here.

Credentials

Only one required env var (PCM_ANY_KEY) is declared, and its use is consistent with non-chat API calls. The SKILL.md explicitly notes chat endpoints must use per-agent keys declared elsewhere, which reduces overbroad credential use.

Persistence & Privilege

Skill is not forced-always, does not request system config paths, and is user-invocable. It does not modify other skills or request elevated system presence.

Assessment

This skill appears to be a straightforward API reference for the Pincaimao platform. Before installing, ensure you trust https://www.pincaimao.com and that the PCM_ANY_KEY you provide has minimal privileges needed (prefer a key scoped to file and non-chat operations). Keep chat-specific keys separate as recommended. Be cautious about uploading sensitive data (resumes, audio, or other files) because the skill's examples send that data to api.pincaimao.com. Because this is instruction-only (no code files), there is nothing hidden in binaries — but also note the agent will send your data and Authorization header to the documented endpoints, so protect and rotate keys as appropriate and avoid placing keys in shared or public environments.

聘才猫平台基础能力

所有聘才猫智能体 skill 共用的底层接口。具体智能体（JD 助手、简历匹配等）的 inputs 参数见各自的 skill，通用部分统一在此。

---

鉴权

接口	Key 要求
chat-messages	必须使用该智能体专属 key，不同智能体 key 不能混用
其他所有接口	任意创建的 key 均可

# chat-messages 示例（key 由各智能体 skill 的 primaryEnv 声明，如 PCM_JD_ASSISTANT_KEY）
-H "Authorization: Bearer <agent-specific-key>"

# 其他接口示例
-H "Authorization: Bearer $PCM_ANY_KEY"   # 任意 key 即可

---

1. chat-messages（智能体对话）

POST https://api.pincaimao.com/agents/v1/chat/chat-messages

请求体

{
  "query": "...",
  "inputs": {},
  "response_mode": "blocking",
  "conversation_id": null,
  "user": null
}

字段	必填	说明
query	是	依智能体而定
inputs	否	各智能体自定义参数
response_mode	否	"blocking" 或 "streaming"，默认 streaming
conversation_id	否	多轮对话时传入，保持上下文
user	否	业务侧用户 ID

Blocking 响应

{
  "event": "message",
  "message_id": "...",
  "conversation_id": "...",
  "answer": "AI 回复内容",
  "metadata": { "usage": { "total_tokens": 953 } }
}

取结果：.answer

Streaming 响应（SSE）

每行以 data: 开头，包含 JSON 对象：

事件	说明	关键字段
message_start	开始生成	conversation_id, message_id
message / agent_message	内容片段	answer（需拼接）
node_finished	节点完成	data.inputs.LLMText（完整文本）, data.title
message_end	生成结束	—

特殊：node_finished 且 data.title 含"聘才猫任务结束"→ 任务完成。

Streaming 解析模板

curl -N -s -X POST 'https://api.pincaimao.com/agents/v1/chat/chat-messages' \
  -H "Authorization: Bearer <agent-specific-key>" \
  -H 'Accept: text/event-stream' \
  -H 'Content-Type: application/json' \
  -d '{"query":"...","inputs":{},"response_mode":"streaming"}' \
| python3 -c "
import sys, json
answer = ''
for line in sys.stdin:
    line = line.strip()
    if not line.startswith('data: '):
        continue
    try:
        d = json.loads(line[6:])
        event = d.get('event', '')
        if event in ('message', 'agent_message'):
            answer += d.get('answer', '')
        elif event == 'message_end':
            break
    except json.JSONDecodeError:
        pass
print(answer)
"

---

2. 文件上传

POST https://api.pincaimao.com/agents/v1/files/upload

curl -s -X POST 'https://api.pincaimao.com/agents/v1/files/upload' \
  -H "Authorization: Bearer $PCM_ANY_KEY" \
  -F "file=@/path/to/file.pdf"

响应：

{
  "filename": "resume.docx",
  "ext": "docx",
  "cos_key": "/resources/file/20250711/abc123.docx",
  "presigned_url": "https://..."
}

注意：后续传给智能体的是 cos_key，不是 presigned_url。支持格式：xlsx、docx、pdf、txt、csv、jpg、png，最大 50MB。

---

3. 获取文件临时链接

GET https://api.pincaimao.com/agents/v1/files/presigned-url

curl -s "https://api.pincaimao.com/agents/v1/files/presigned-url?cos_key=/resources/file/xxx/abc.jpg" \
  -H "Authorization: Bearer $PCM_ANY_KEY"

响应包含 file_url（10 分钟有效）。可选参数 expired（秒，最长 30 天）。

---

4. 获取会话列表

GET https://api.pincaimao.com/agents/v1/chat/conversations

curl -s "https://api.pincaimao.com/agents/v1/chat/conversations?limit=20&user=user_001" \
  -H "Authorization: Bearer $PCM_ANY_KEY"

参数	说明
last_id	游标分页，上一页最后一条 ID
user	按用户 ID 筛选
limit	每页条数，默认 20，最大 100

响应：data.data[] 含 id（conversation_id）、name、inputs、created_at。

---

5. 获取会话历史消息

GET https://api.pincaimao.com/agents/v1/chat/messages

curl -s "https://api.pincaimao.com/agents/v1/chat/messages?conversation_id=xxxx&limit=20" \
  -H "Authorization: Bearer $PCM_ANY_KEY"

倒序返回。参数 first_id 用于翻页。响应 data.data[] 含 query、answer、inputs、created_at。

---

6. 音视频转文字

POST https://api.pincaimao.com/agents/v1/tts/audio_to_text

curl -s -X POST 'https://api.pincaimao.com/agents/v1/tts/audio_to_text' \
  -H "Authorization: Bearer $PCM_ANY_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"cos_key": "/resources/file/xxx/audio.mp3"}'

响应：data.text（转写文字）。目前仅支持 cos_key 路径。

---

7. 简历 JSON 上传

POST https://api.pincaimao.com/v1/files/resume-json

将结构化简历数据上传，返回 cos_key 供后续智能体使用。

curl -s -X POST 'https://api.pincaimao.com/v1/files/resume-json' \
  -H "Authorization: Bearer $PCM_ANY_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "name": "张三",
    "gender": "男",
    "age": "28",
    "degree": "本科",
    "work_year": "5",
    "city": "广州市",
    "phone": "13800138000",
    "email": "[e-mail]",
    "apply_job": "软件工程师",
    "education_objs": [...],
    "job_exp_objs": [...],
    "skills_objs": [{"skills_name": "Java", "skills_level": "精通"}]
  }'

响应：data.cos_key（供后续接口使用）。

<details> <summary>简历 JSON 完整字段列表</summary>

基本信息：name, gender, age, marital_status, birthday, race, hometown_address, nationality, polit_status, city, phone, email, qq, weixin, avatar_url, avatar_data

求职意向：apply_job, apply_cpy, work_year, work_position, work_company, work_status, work_salary, work_job_nature

学历：college, college_dept, major, degree, recruit

自述：cont_my_desc, cont_hobby

期望工作 (expect_job_objs[])：expect_job, expect_cpy, expect_salary, expect_industry, expect_time, expect_jnature, expect_jstatus, expect_jlocation

教育经历 (education_objs[])：start_date, end_date, edu_college, edu_college_dept, edu_major, edu_recruit, edu_degree, edu_content

工作经历 (job_exp_objs[])：start_date, end_date, job_cpy, job_position, job_dept, job_nature, job_salary, job_location, job_content

项目经历 (proj_exp_objs[])：start_date, end_date, proj_name, proj_cpy, proj_position, proj_content, proj_resp

培训经历 (training_objs[])：start_date, end_date, train_org, train_loc, train_name, train_cert, train_cont

技能 (skills_objs[])：skills_name, skills_level（掌握/熟悉/擅长/精通）

语言 (lang_objs[])：language_name, language_level

证书奖项 (all_cert_objs[])：cert_name, cert_type（certificate 或 award）

</details>

---

常见错误

问题	原因	解决
401	Key 错误或混用	chat-messages 必须用智能体专属 key
上传返回非 200	文件超 50MB 或格式不支持	检查文件大小和格式
Streaming 无输出	curl 缺 -N 标志	加 -N 禁用缓冲
临时链接失效	presigned_url 超过 10 分钟	重新调用 presigned-url 接口

---

External Endpoints

• https://api.pincaimao.com — Pincaimao platform API (chat, file upload, conversations)

Security & Privacy

• API key is read from environment variable and passed via Authorization header; never hardcoded
• Resume files, job descriptions, and contract text are transmitted to api.pincaimao.com for AI processing
• Uploaded files are stored on Pincaimao's COS (Cloud Object Storage); returned cos_key paths should be treated as sensitive
• This skill does not store, log, or transmit data to any endpoint other than api.pincaimao.com
• Safe to invoke autonomously; all network calls are scoped to the authenticated user's API key