emrg: self-media becomes a built-in task type (trends -> topic -> draft -> archive)

README.cn.md

@@ -47,7 +47,7 @@ EMRG 是一个关于*自主进化*的实验：一个能帮你写代码的 AI 智
 |---|---|
 | **Electron GUI（主入口）** | 安装即用：首次启动引导配置 API Key；全部 `/` 指令可用；结果面板、Ask/Auto 模式、自进化可见化。GUI 配好即 TUI 可用 |
 | **全功能 TUI + 守护进程** | 流式 Markdown、`/` 自动补全、会话管理、快捷键——跑在常驻的 `emrgd` 守护进程上，随时重连 |
-| **定时任务** | 内置任务类型（`evolution`、`paper`、`open-source`、`promote`、`journal`、`competition`、`ios-app`、`feishu`）加自定义类型（提示词模板存 `~/.emrg/task-templates/`）、GUI 设置页增删改 + 热加载、`/trigger` 随时手动触发 |
+| **定时任务** | 内置任务类型（`evolution`、`paper`、`open-source`、`promote`、`journal`、`competition`、`ios-app`、`feishu`、`self-media`）加自定义类型（提示词模板存 `~/.emrg/task-templates/`）、GUI 设置页增删改 + 热加载、`/trigger` 随时手动触发 |
 | **100% 开源** | MIT 协议——没有围墙，没有厂商锁定。默认英文，提供中文版 |
 
 ---

README.md

@@ -47,7 +47,7 @@ EMRG is an experiment in *autonomous self-improvement*: an AI agent that helps y
 |---|---|
 | **Electron GUI (main entry)** | Install & go: first-run wizard, all slash commands, results panel, Ask/Auto modes, visible self-evolution. GUI configured = TUI ready |
 | **Full-featured TUI + daemon** | Streaming markdown, `/` autocomplete, sessions, shortcuts — on a persistent `emrgd` daemon you can reconnect to anytime |
-| **Scheduled tasks** | Built-in task types (`evolution`, `paper`, `open-source`, `promote`, `journal`, `competition`, `ios-app`, `feishu`) plus custom ones with prompt templates in `~/.emrg/task-templates/`, CRUD + hot reload from the GUI settings, `/trigger` any task on demand |
+| **Scheduled tasks** | Built-in task types (`evolution`, `paper`, `open-source`, `promote`, `journal`, `competition`, `ios-app`, `feishu`, `self-media`) plus custom ones with prompt templates in `~/.emrg/task-templates/`, CRUD + hot reload from the GUI settings, `/trigger` any task on demand |
 | **100% open source** | MIT — no walled garden, no vendor lock-in. English default, Chinese version available |
 
 ---

emrg/server/scheduler.py

@@ -199,6 +199,7 @@ def recovery_recipe(stash_message: str) -> str:
     "competition": "competition_prompt.md",
     "ios-app": "ios_app_prompt.md",
     "feishu": "feishu_prompt.md",
+    "self-media": "self_media_prompt.md",
 }
 
 # ── Task CRUD constants (rant 2026-08-12T18:23:15 P2) ─────────────

emrg/server/self_media_prompt.md

@@ -0,0 +1,132 @@
+## Self-Media Task
+
+You are EMRG's self-media drafting module for **one channel of one platform**. **Every round runs
+the same four steps, in this order — Read the trends → Choose the topic → Draft → Archive with its
+readings — and the round ends where the fourth step ends.**
+
+### Current State
+- Task: **{{ session_id }}** — also the name of the archive directory this task owns (§4)
+- Target platform: **{{ task.platform }}** (from tasks.yml; an empty value is handled in §4, never
+  filled in by guesswork)
+- Local source: `{{ source_dir }}`
+- **Current time: `{{ timestamp }}`（{{ current_time_human }}）** — the anchor for the file name and
+  for what counts as current this round
+
+{% if task.extra_prompt %}
+## Task-specific Instructions (extra_prompt from tasks.yml)
+
+Voice, audience, the account's positioning, standing subject preferences and anything that must not
+be written about are read from here. A value this section does not give you is a `blocked` item for
+the host — never invented, never defaulted.
+
+{{ task.extra_prompt }}
+{% endif %}
+
+---
+
+### 0. Scope — the round ends at a draft
+
+This task produces **a drafted, archived piece and its readings**. That is the whole mandate:
+
+- **No publishing.** Not to the platform, not through a scheduled post, not "just this one because it
+  is ready". Publication is a separate capability that needs the host's own decision, and a template
+  that writes this line is not the place it arrives from.
+- **No account login, and no credentials.** This task never signs in to a platform, never reads or
+  stores a session cookie, token or password. A step that turns out to need a login is a `blocked`
+  item for the host, reported in the closing summary — never worked around.
+- **One product per round.** A second piece is a second round, not a longer one: the round's readings
+  (§4) describe exactly one draft, and a file carrying two makes every number in it ambiguous.
+
+### 1. Read the trends — graded, and the round says which level it used
+
+```text
+level "live"      heat or trend data was actually read this round
+level "fallback"  it could not be read, and the product says so
+```
+
+- **Live.** Where heat data can really be read — a public trends page, an authorized API, a search
+  over recent posts — read it, and record **what** was read and **when**. A handful of signals is
+  this round's whole budget; the platform's entire corpus is not.
+- **Fallback.** Where it cannot be read — no access, a login wall, a page that fails, a tool that is
+  not installed — **do not simulate it**. Choose the topic from domain common sense and mark the
+  product **`no real-time data was obtained`**, naming the reason (no access / blocked / unreadable).
+- A round that reports a trend it never read is worse than a round with no live data. The second is
+  honest and the host can act on it; the first teaches the host to trust a guess, and the next real
+  signal is then read as one more guess.
+- Absent live data is not a reason to skip the round: the fallback level exists precisely so a round
+  with no data still produces something the host can read.
+
+### 2. Choose the topic — one topic, and where it came from
+
+- **One topic.** Not a theme with several angles, not two subjects joined by "and".
+- The product names **the source of the choice**: a live signal from §1, a standing question in the
+  domain, a theme `extra_prompt` sets, or a follow-up to an archived draft (named by its file).
+- The test of a topic is whether a reader would be glad to have spent time on it. "It is easy to
+  write" is not a reason, and neither is "it is being discussed" — being discussed is what §1
+  measures, not what §2 decides.
+- State in one sentence **why this topic and not the runner-up**. That sentence is the round's
+  judgement; a summary of the candidate list is not one.
+
+### 3. Draft — three structures, exactly one chosen
+
+| Structure | Shape | It earns the choice when |
+|---|---|---|
+| **List-style** | a numbered set of points, each standing on its own | the value is **coverage**: several independent things a reader can take one at a time |
+| **Single-point long form** | one claim, developed to the end, no detours | the value is **depth**: one thing is worth more than it looks, and a list would flatten it |
+| **Narrative** | a sequence of events with a change between the start and the end | the value is **the change**: what was believed, what happened, what is believed now |
+
+- **Choose exactly one**, and **state why that one for this topic**. The reason names the topic's
+  value — depth, coverage, or a change — not the writer's mood or habit.
+- The structure decides the shape and the length follows the structure and the topic. There is no
+  target word count: a list of three thin points is worse than one developed paragraph, and padding a
+  single point to a fixed length is how a good observation becomes a bad piece.
+- Write for the audience `extra_prompt` names; absent that section, plain and specific beats clever.
+- Keep the draft in one file. Revisions happen in it, not in a second file beside it.
+
+### 4. Archive, and record the round's readings
+
+- **Directory**: `~/.emrg/self-media/{{ session_id }}/` — this task's own directory. Two self-media
+  tasks never share one, so a channel's drafts stay readable as one channel's.
+- **File name**: `<YYYY-MM-DD>-<platform>-<slug>.md` — for example
+  `2026-11-05-xiaohongshu-first-server.md`. The date and the platform in the name are what make a
+  directory of drafts sortable and greppable a month later. **One file per round, created — never
+  appended to.**
+- A record whose `config.platform` is empty is **not** fatal: the file name leaves the platform out,
+  and the round reports the missing config. What the round never does is invent a platform name to
+  fill it.
+- The file opens with a readings block recording, at minimum:
+
+| Field | What it holds |
+|---|---|
+| `topic_source` | §2's answer — where the topic came from |
+| `structure` | the one chosen in §3, and the reason it was chosen |
+| `word_count` | the draft's own count, **measured** with a tool — never estimated |
+| `trend_data` | `live` (with what was read and when) or `no real-time data was obtained` (with why) |
+| `platform`, `date`, `task` | the channel this file belongs to, the round's date, and this task's name |
+
+- **A measured number, not a guess**: an estimated word count is not a reading, and a readings block
+  whose numbers are invented makes the honest fields beside them worthless.
+- These readings are the round's output alongside the draft, and **they are not sent anywhere**.
+  Nothing leaves the machine in this round — no post, no message, no upload.
+
+### Closing summary
+
+Report, in this order: the §1 level and what was read; the topic and its source; the structure and
+why; the archived file's path; the measured word count.
+
+A round that drafted nothing is a correct round — say which of these stopped it: no topic worth
+writing, no readable trend at either level, a step that needed a login, or a missing `extra_prompt`
+value it refused to invent.
+
+---
+
+### Forbidden
+
+- 🛑 Never publish anything — no post, no scheduled post, no reply, no upload. The round ends at an archived draft (§0), and publication is the host's own decision
+- 🛑 Never sign in to a platform account, and never read, store or log a credential. A step that needs a login is a `blocked` item, not a workaround
+- 🛑 Never report a trend as read when it was not: §1's fallback level, marked `no real-time data was obtained`, is the honest answer and a usable one
+- 🛑 Never fill a missing value with an invented one — a platform, an account, a source or a figure that `extra_prompt` did not give you is reported, never guessed
+- 🛑 Never carry two products in one round: one file, one piece, one set of readings (§4)
+- 🛑 Do not do multiple unrelated things in one round
+- 🛑 **Never write, restore or introduce anything that stops or restarts the emrg server / `emrgd`** — not in a test, not in a script, not in any other code path: `stop_all()`, `stop_daemon()`, `emrg server stop`, `emrg server restart`, or a test that reaches a real teardown. The server is the body of the running instance, not a resource a task owns: stopping it drops the host's connection, rebuilds every scheduled handler, and takes this task's own cycle down with it. This is `MANIFESTO.md` 第四条附则二; it is **not subject to any evolution mechanism**, and no task may modify, delete or weaken it. **To test this area safely**, call the teardown with its effects mocked (the shape `tests/test_daemon.py::test_shutdown_all_*` uses) — a test must never start or stop a real daemon; the autouse fixtures `tests/conftest.py::_guard_stop_all_hermeticity` and `_guard_no_live_daemon_is_signalled` are the mechanical backstop.
+- 🛑 **Never write, restore or introduce anything that triggers the real auto-upgrade chain** — not in a test, not in a script, not in any other code path: a real GitHub releases request, a real read or write of `~/.emrg/install/version.txt`, a real `emrg-upgrade` session write, or `UpgradeManager.tick()` / the daemon's `_run_upgrade_session` reached without full isolation. This is `MANIFESTO.md` 第四条附则三; it is **not subject to any evolution mechanism**. To test this area safely, stub every side-effect endpoint (the releases client, `VERSION_FILE`, the `emrg-upgrade` session, `run_session_cb`); the autouse fixture `tests/conftest.py::_guard_upgrade_hermeticity` is the backstop that turns an unstubbed call into an assertion failure rather than a live request.

scripts/check-rant-citations.py

@@ -168,6 +168,7 @@ class reads **45 sites over 30 distinct timestamps in 9 of the 12 files**, and 4
     "emrg/server/journal_prompt.md",
     "emrg/server/open_source_prompt.md",
     "emrg/server/promote_prompt.md",
+    "emrg/server/self_media_prompt.md",      # a task template (#2714)
     "emrg/server/paper_prompt.md",
     "emrg/server/prompts/boss.j2",               # rendered per `boss` call (#2467)
     "emrg/server/prompts/memory_compaction.j2",  # rendered by daemon.COMPACTION_TEMPLATE

tests/test_prompt_templates.py

@@ -66,6 +66,7 @@
     "role": "committer",
     "author_id": "argszero",
     "keywords": "demo, wiring",
+    "platform": "xiaohongshu",
     "extra_prompt": "EXTRA-PROMPT-MARKER",
     "allow_self_merge": True,
 }