diff --git a/.gitignore b/.gitignore
index 21f7494d..0559b2bc 100644
--- a/.gitignore
+++ b/.gitignore
@@ -78,6 +78,9 @@ memory/L4_raw_sessions/*
 # Plan SOP
 !memory/plan_sop.md
 
+# UltraPlan SOP
+!memory/ultraplan_sop.md
+
 # Goal Mode SOP
 !memory/goal_mode_sop.md
 
diff --git a/assets/ga_httpapp.py b/assets/ga_httpapp.py
new file mode 100644
index 00000000..29026877
--- /dev/null
+++ b/assets/ga_httpapp.py
@@ -0,0 +1,51 @@
+import threading, sys, os
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+from fastapi import FastAPI, Header, HTTPException, Query
+from pydantic import BaseModel
+from agentmain import GeneraticAgent
+
+PORT, API_KEY = int(sys.argv[1]), sys.argv[2]
+app, agent, lock = FastAPI(), GeneraticAgent(), threading.Lock()
+task_id, outputs, stopped, base = 0, [], True, 0
+threading.Thread(target=agent.run, daemon=True).start()
+class Req(BaseModel): prompt: str
+agent.verbose = False
+
+def auth(key):
+    if key != API_KEY: raise HTTPException(404)
+
+def run_task(tid, prompt):
+    global stopped
+    try:
+        dq = agent.put_task(prompt, source="http")
+        while "done" not in (item := dq.get(timeout=1200)):
+            if "next" in item:
+                t = base + item.get("turn", 0)
+                while len(outputs) <= t: outputs.append({"task_id": tid, "text": ""})
+                outputs[t] = {"task_id": tid, "text": item["outputs"][-1] if item.get("outputs") else ""}
+        for i, txt in enumerate(item.get("outputs", [])):
+            t = base + i
+            while len(outputs) <= t: outputs.append({"task_id": tid, "text": ""})
+            outputs[t] = {"task_id": tid, "text": txt}
+    finally: stopped = True
+
+@app.post("/put_task")
+def put_task(req: Req, key: str = Header(alias="X-API-Key")):
+    auth(key); global task_id, stopped, base
+    with lock:
+        if not stopped: return {"ok": False, "error": "should abort first"}
+        stopped = False; task_id += 1; base = len(outputs)
+    threading.Thread(target=run_task, args=(task_id, req.prompt), daemon=True).start()
+    return {"ok": True, "task_id": task_id}
+
+@app.post("/abort")
+def abort(key: str = Header(alias="X-API-Key")): auth(key); agent.abort(); return {"ok": True}
+
+@app.get("/output")
+def get_output(key: str = Header(alias="X-API-Key"), k: int = Query(5)):
+    auth(key)
+    recent = [o["text"] for o in outputs[-k:]] if outputs else []
+    return {"task_id": task_id, "stopped": stopped, "output": "\n".join(recent)}
+
+if __name__ == "__main__":
+    import uvicorn; uvicorn.run(app, host="0.0.0.0", port=PORT)
diff --git a/assets/ga_ultraplan.py b/assets/ga_ultraplan.py
index ee920b1a..06e4f90d 100644
--- a/assets/ga_ultraplan.py
+++ b/assets/ga_ultraplan.py
@@ -4,9 +4,9 @@ from time import time, sleep
 import html, os, subprocess, sys, tempfile, threading, webbrowser
 from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 
-__all__ = ["phase", "parallel", "pipeline"]
+__all__ = ["phase", "parallel", "mapchain"]
 
-_T0 = time(); _phases = []; _phase_stack = []; _tasks = []; _current = "idle"; _events = []; _srv = None; _last = time(); _lock = threading.Lock()
+_T0 = time(); _RUN_DIR = None; _phases = []; _phase_stack = []; _tasks = []; _current = "idle"; _events = []; _srv = None; _last = time(); _lock = threading.Lock()
 
 def _note(s):
     global _last
@@ -91,9 +91,12 @@ def _fmt(x, data):
     return x.format(**data) if isinstance(x, str) else x
 
 def _subagent(desc, prompt=None, *, llm_no=0, timeout=3600):
+    global _RUN_DIR
     root = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
     os.makedirs(tmp := os.path.join(root, "temp"), exist_ok=True)
-    with tempfile.NamedTemporaryFile("w", encoding="utf-8", suffix=".txt", prefix="ultra_", dir=tmp, delete=False) as f:
+    if _RUN_DIR is None:
+        _RUN_DIR = os.path.join(tmp, f"ultraplan_{int(_T0)}_{os.getpid()}"); os.makedirs(_RUN_DIR, exist_ok=True)
+    with tempfile.NamedTemporaryFile("w", encoding="utf-8", suffix=".txt", prefix="ultra_", dir=_RUN_DIR, delete=False) as f:
         f.write(desc if prompt is None else prompt)
     print(f"[subagent] {desc} -> {f.name}", flush=True); _note(f"agent: {desc}")
     cmd = [sys.executable, os.path.join(root, "agentmain.py"), "--func", f.name, "--llm_no", str(llm_no), "--nobg", "--nolog"]
@@ -125,9 +128,9 @@ def parallel(tasks, max_workers=None, _label=None, **data):
     with ThreadPoolExecutor(max_workers=max_workers or len(tasks) or 1) as ex:
         return list(ex.map(lambda t: _run(t, data), tasks))
 
-def pipeline(items, *steps, max_workers=None, **data):
+def mapchain(items, *steps, max_workers=None, **data):
     global _current
-    items = list(items); label = f"pipeline: {len(items)} items x {len(steps)} steps"
+    items = list(items); label = f"mapchain: {len(items)} items x {len(steps)} steps"
     with _lock: _current = label
     def run(x):
         for step in steps:
diff --git a/memory/incubator_sop.md b/memory/incubator_sop.md
index d85c1c4f..00e18f20 100644
--- a/memory/incubator_sop.md
+++ b/memory/incubator_sop.md
@@ -15,16 +15,15 @@ memory下大部分文件不适合分发，不要复制 memory 下未被 gitignor
 - `GARoot/*.py` 必须包含根目录所有 `.py`。
 - `GARoot/assets/*.txt *.json` 必须包含 assets 顶层所有 `.txt`/`.json`。
 - `GARoot/memory/` 只取 `.gitignore` 白名单/已允许分发文件；排除 `global_mem.txt`、`global_mem_insight.txt`、`__pycache__/`、`*.pyc`。
-- 按当前清单实测压缩包约153KB/55文件；正常不应超过170KB，文件数不应超过60。
+- 按当前清单实测压缩包约153KB/55文件；正常不应超过200KB，文件数不应超过60。
 
 ## 依赖
 requests beautifulsoup4
 尽量复用远端已有python/venv
 
 ## 通信
-看subagent.md
-同subagent协议：`agentmain.py --task {name} --input "..."`
-或起reflect worker并设置bbs信息
+1. **首选** 阅读 `assets/ga_httpapp.py`（HTTP API，~50行自解释）
+2. 备选：subagent.md 文件协议 或 reflect worker + bbs
 
 ## 干预记忆
 直接编辑远端 memory/ 下的文件（SOP/全局记忆）
diff --git a/memory/ultraplan_sop.md b/memory/ultraplan_sop.md
new file mode 100644
index 00000000..a708c708
--- /dev/null
+++ b/memory/ultraplan_sop.md
@@ -0,0 +1,428 @@
+# GA UltraPlan SOP
+
+UltraPlan is GA's controller-driven protocol for work that is too broad, adversarial, or uncertain for a single linear agent pass.
+
+The core model is **not** "one script equals one MapReduce". A script is a **bounded, file-backed MapReduce program**: it may contain multiple phases, multiple fan-out/fan-in passes, and multiple per-item chains, but it should only cover the portion of the task that is currently plannable. When the script finishes, UltraPlan is not automatically finished. The main agent must read the outputs, reduce them into task state, and decide whether to stop, act directly, ask the user, or launch the next UltraPlan script.
+
+```text
+main agent controller
+  -> write/run a bounded UltraPlan script
+  -> read file-backed outputs
+  -> reduce into shared state / decision
+  -> stop, direct edit, ask, or write the next script
+```
+
+## Non-negotiable principles
+
+1. **Script completion is not task completion.** All phases being done only means the currently planned program returned evidence.
+2. **Plan only the currently plannable horizon.** If later work depends heavily on unknown results, stop after producing those results and let the main agent regenerate the next script.
+3. **Everything important is file-backed.** Subagents return `.out.txt` files; shared state lives in explicit files under the current working directory, usually `./temp/ultraplan_<task>/`.
+4. **The main agent owns global reduction.** Subagents produce local evidence, attempts, critiques, or transforms. The main agent owns cross-run state, prioritization, final decisions, and user-facing claims.
+5. **Use independent perspectives on purpose.** Parallel agents should not be redundant copies unless redundancy itself is the experiment.
+6. **No silent completeness claims.** If coverage is bounded, sampled, timed out, or intentionally skipped, say so.
+
+## Workflow script contract
+
+A GA UltraPlan script is plain Python using `../assets/ga_ultraplan.py`:
+
+```python
+from assets.ga_ultraplan import phase, parallel, mapchain
+```
+
+Run scripts from the GA code root so `assets` imports correctly. Express orchestration only through `phase()`, `parallel()`, and `mapchain()`; design prompts, shared files, and reduction steps around those primitives.
+
+Each task can be one of:
+
+```python
+("Short display name", "Prompt template with {item}, {previous}, {state_dir}, ...")
+
+{
+    "desc": "Short display name with {item}",
+    "prompt": "Prompt template",
+    "data": {"extra": "values"},
+    "llm_no": 0,
+    "timeout": 3600,
+}
+
+lambda: ("Short display name", "Prompt generated lazily at runtime")
+```
+
+A task returns the path of its subagent output file, normally `.../temp/ultra_xxx.out.txt`. Treat that path as the workflow value. Do not rely on subagents producing valid JSON unless a later Python step explicitly validates and repairs it.
+
+## Shared workspace: give independent agents just enough shared context
+
+Subagents are independent. A prompt alone is often too narrow: it lacks the evolving global view and encourages duplicated work. For any serious UltraPlan run, create only the shared files this task needs, under the current working directory, usually `./temp/ultraplan_<target>/`.
+
+Do not impose a fixed schema. Use task-shaped files such as `sop_context.md`, `current_findings.md`, `candidate_urls.txt`, `failed_attempts.md`, `coverage_list.txt`, or `acceptance_criteria.md`. One context file may be enough; a Sweep may need an item list; a Hunt may need prior attempts; an Improve run may need the target plus criteria.
+
+Rules:
+
+- The controller creates and refreshes shared files before launching a script.
+- Worker prompts say exactly which shared file(s) to read and which target item, lens, or shard they own.
+- Workers may write scratch notes only when explicitly assigned a unique file. Avoid concurrent appends to the same file.
+- Reducers read worker `.out.txt` files plus the relevant shared files; they do not infer global truth from one worker.
+- Between scripts, the main agent updates whichever notes are useful: tried attempts, accepted findings, rejected claims, remaining frontier, coverage bounds, or changed priorities.
+
+## Primitive semantics
+
+### `phase(name, desc="")`
+
+Use `phase` as a visible boundary in the HTML phase tree and as a semantic boundary in the program. A phase should describe what is being learned, attempted, or reduced.
+
+```python
+with phase("Explore target surface", "map unknown areas before choosing attacks"):
+    maps = parallel([...])
+```
+
+Good phase names are verbs tied to the archetype: `Hunt candidates`, `Verify hits`, `Critique improvement angles`, `Sweep item list`, `Reduce frontier`.
+
+### `parallel(tasks, max_workers=None, **data)`
+
+Run independent tasks concurrently and return output-file paths in input order. This is a barrier: the next Python statement runs only after all tasks complete.
+
+Use `parallel` when a reduction truly needs the full set:
+
+- brainstorming diverse candidates before choosing the best;
+- independent sharp critics attacking the same target from different angles;
+- partitioned sweep results that must be deduplicated;
+- a judge/reducer comparing mutually exclusive approaches.
+
+Do not insert a barrier merely because stages have different names. If each item can move through its stages independently, use `mapchain`.
+
+### `mapchain(items, *steps, max_workers=None, **data)`
+
+Run each item through all steps independently. There is no cross-item barrier between stages: item A can be in verification while item B is still being inspected.
+
+Each step receives the current value as `{item}` and `{previous}`. Initially this is the original item; after each subagent task it becomes that subagent's `.out.txt` path.
+
+When a later step consumes a prior `.out.txt`, instruct it to **tail** that file, not read it, unless the full transcript matters. `.out.txt` files contain the previous worker's full transcript; the useful conclusion is usually near the end.
+
+```python
+context_file = f"{state_dir}/context.md"
+reports = mapchain(
+    items,
+    ("Inspect {item}", f"Read {context_file} and inspect assigned item: {{item}}. Output evidence-backed findings."),
+    ("Verify {previous}", "Tail {previous}. Try to refute or reproduce the finding with tools."),
+    state_dir=state_dir,
+)
+```
+
+Use `mapchain` by default for per-item multi-stage work. It keeps item context local, avoids artificial barriers, and prevents the main agent from hand-managing intermediate maps.
+
+## Controller cycle: UltraPlan is recursive MapReduce, not a single run
+
+A full UltraPlan often needs several scripts. Each script is one bounded program over the currently knowable horizon; the main agent is the controller between programs.
+
+Use this cycle:
+
+1. **Scout cheaply in the main agent.** Read the request, obvious files, target pages, logs, or prior notes. Do not spawn blind fan-out before identifying the target object.
+2. **Create or refresh only the needed shared notes.** Use task-shaped files such as a target/context note, criteria, attempt history, item list, or current findings. Do not assume a universal schema.
+3. **Choose the current archetype.** Hunt, Improve, Explore, or Sweep. A script may contain multiple phases and multiple MapReduce barriers, but it should have one dominant purpose.
+4. **Write and run the bounded script.** Encode the control flow up front using `phase`, `parallel`, and `mapchain`.
+5. **Read outputs physically.** Read the final reducer and any important intermediate `.out.txt` files. Do not trust file names, status lines, or success messages.
+6. **Reduce into controller state.** Update the useful shared notes: what was tried, what survived verification, what remains unknown, what coverage is bounded, and what changed priority.
+7. **Decide next action.** Stop and answer, edit directly, ask the user, or launch a new UltraPlan script with a new archetype.
+
+This is the key distinction: all phases in one script being complete only ends that script's MapReduce program. It does not prove the overall task is complete.
+
+## Four recursive archetypes
+
+These are not a fixed catalog of examples. They are four first-principles control shapes that can transform into one another. A serious task often moves from one archetype to another across scripts.
+
+### 1. Hunt: goal-directed search until one viable success is found
+
+Use when the task has a clear success condition and many possible routes. You need broad ideation, non-duplicate attempts, and fast validation.
+
+Examples: find a working exploit path, find any viable implementation approach, find one source that proves a claim, find a way around an integration blocker.
+
+Control shape:
+
+1. Expand the candidate frontier from multiple independent lenses.
+2. Deduplicate candidates against the recorded attempt history or other relevant shared notes.
+3. Try candidates in parallel or with `mapchain(candidate -> attempt -> verify)`.
+4. Stop early only if success is physically verified; otherwise update the frontier and launch the next hunt wave.
+
+Prompt requirements:
+
+- Each hunter must state its lens and avoid repeating recorded attempts.
+- Each attempt must specify exact success evidence.
+- Verifiers should default to "not proven" unless they can reproduce or directly inspect the evidence.
+
+### 2. Improve: optimize a target by finding sharp defects or upgrade angles
+
+Use when an object exists and the goal is to make it better. Independence is critical: issue-finders must be pointed, skeptical, and non-colluding, not generic reviewers.
+
+Examples: improve a design, harden a plan, optimize prompts, reduce latency, refine an SOP, improve UX, raise code quality.
+
+Control shape:
+
+1. Freeze the current target object in a concrete shared note or file reference.
+2. Launch independent critics with distinct sharp lenses: correctness, missing cases, simplicity, performance, security, user value, maintainability, originality, verification.
+3. Reduce critiques into a prioritized issue list with evidence.
+4. Launch a second script to sweep fixes or design specific remedies.
+5. Verify that each accepted improvement actually improves the target and does not break constraints.
+
+Prompt requirements:
+
+- Critics should be adversarial and specific, not balanced reviewers.
+- A critique is useful only if it names the exact target fragment, failure mode, and improvement direction.
+- The reducer must reject duplicate, vague, or unsupported criticism.
+
+### 3. Explore: determine what exists, what matters, or whether the target is real
+
+Use when the objective itself is uncertain: you do not yet know the landscape, relevant files, possible mechanisms, or whether a phenomenon exists.
+
+Examples: understand an unfamiliar codebase area, investigate a bug class, discover whether a site exposes an API, map a research topic, determine whether a feature already exists.
+
+Control shape:
+
+1. Partition the unknown space into independent lenses or regions.
+2. Have explorers gather evidence, not final answers.
+3. Reduce into a map of knowns, unknowns, contradictions, and promising next questions.
+4. Convert the next step into Hunt, Improve, or Sweep once the target becomes concrete.
+
+Prompt requirements:
+
+- Explorers must report scope covered and scope not covered.
+- Reducers must distinguish evidence, inference, and speculation.
+- Exploration should produce a frontier, not pretend to close the task.
+
+### 4. Sweep: enumerate and process every item without duplication or omission
+
+Use when the main risk is coverage: every file, row, endpoint, finding, document, or case must be considered.
+
+Examples: per-file migration planning, verify every reported issue, process every source, audit every route, update every usage site.
+
+Control shape:
+
+1. Build an item list and validate the enumeration method.
+2. Use `mapchain` for per-item work: inspect -> transform/plan -> verify.
+3. Reduce item outputs into a coverage table.
+4. Run a gap-checker against the item list and any relevant attempt/coverage notes.
+5. If gaps remain, launch another sweep over the missing set.
+
+Prompt requirements:
+
+- Each worker owns exactly one item or shard.
+- Each output must include item ID, action taken/proposed, evidence, and unresolved risk.
+- Reducers must report coverage count and omissions explicitly.
+
+## Archetype transformations
+
+The four archetypes recursively call each other:
+
+- An **Improve** run may discover unknown areas, causing an **Explore** sub-run.
+- An **Explore** run may identify a success route, causing a **Hunt**.
+- A **Hunt** that finds a candidate may require **Improve** to harden it.
+- Any accepted plan may require **Sweep** to apply or verify it across every item.
+- A **Sweep** may expose failures, returning to **Hunt** for workarounds or **Improve** for fixes.
+
+Do not force all transformations into one script. If the next archetype depends on the reducer's result, stop the script, update shared state, and write a new script.
+
+## Script templates by archetype
+
+The examples below are starting points. Replace placeholders with concrete shared files and target objects. Keep scripts short enough that the controller can read outputs and decide the next archetype.
+
+### Hunt template
+
+```python
+# Run from GA code root.
+from assets.ga_ultraplan import phase, parallel, mapchain
+
+state_dir = "./temp/ultraplan_<target>"
+context_file = f"{state_dir}/context.md"      # task-shaped target/context note
+attempts_file = f"{state_dir}/attempts.md"    # tried candidates and rejected repeats
+criteria_file = f"{state_dir}/criteria.md"    # success evidence or constraints, if useful
+
+with phase("Expand candidate frontier", "independent lenses, avoid recorded duplicates"):
+    candidate_notes = parallel([
+        ("Known-pattern hunter", f"Read {context_file} and {attempts_file}. Propose candidates from known patterns only; exclude repeats."),
+        ("Weird-angle hunter", f"Read {context_file} and {attempts_file}. Propose unconventional candidates; explain why each might work."),
+        ("Constraint hunter", f"Read {context_file} and {criteria_file}. Propose candidates that exploit constraints or narrow success criteria."),
+    ], state_dir=state_dir, max_workers=3)
+
+with phase("Reduce candidates", "dedupe and select attempts"):
+    shortlist = parallel([
+        ("Candidate reducer", f"Read {attempts_file} and these outputs: {candidate_notes}. Write a deduplicated attempt list with success evidence for each.")
+    ])[0]
+
+with phase("Attempt and verify", "each candidate advances independently"):
+    # In real use, the controller may parse/validate shortlist into candidate IDs before this phase.
+    results = parallel([
+        ("Attempt selected candidates", f"Read {shortlist}. Try the strongest candidates. For each, report physical evidence of success or failure.")
+    ])
+
+print(results)
+```
+
+### Improve template
+
+```python
+from assets.ga_ultraplan import phase, parallel
+
+state_dir = "./temp/ultraplan_<target>"
+target_file = f"{state_dir}/target_or_draft.md"
+criteria_file = f"{state_dir}/criteria.md"          # optional constraints/current requirements
+findings_file = f"{state_dir}/current_findings.md"  # optional known claims or prior verification notes
+
+with phase("Sharp independent critique", "specific failure modes, not generic review"):
+    critiques = parallel([
+        ("Correctness critic", f"Read {target_file} and {criteria_file}. Find exact correctness failures; cite target fragments."),
+        ("Simplicity critic", f"Read {target_file}. Find over-complex or confused parts; propose simpler structure."),
+        ("Verification critic", f"Read {target_file} and {findings_file}. Find claims or changes that lack executable verification."),
+        ("Originality critic", f"Read {target_file}. Identify derivative, generic, or weak ideas; propose stronger framing."),
+    ], state_dir=state_dir, max_workers=4)
+
+with phase("Prioritize improvements", "reject vague or duplicate criticism"):
+    issue_list = parallel([
+        ("Issue reducer", f"Read critiques: {critiques}. Produce prioritized issues with exact target location, evidence, and fix direction. Reject duplicates and vague comments.")
+    ])[0]
+
+print(issue_list)
+```
+
+Usually the next script after this template is a **Sweep** over accepted issues or target sections.
+
+### Explore template
+
+```python
+from assets.ga_ultraplan import phase, parallel
+
+state_dir = "./temp/ultraplan_<target>"
+context_file = f"{state_dir}/context.md"          # what is known about the unknown space
+findings_file = f"{state_dir}/current_findings.md" # optional assumptions or prior evidence
+
+with phase("Map unknown space", "evidence first, no premature final answer"):
+    maps = parallel([
+        ("Surface mapper", f"Read {context_file}. Enumerate visible components/files/endpoints/sources and what each proves."),
+        ("Mechanism mapper", f"Read {context_file}. Investigate how the relevant mechanism likely works; separate evidence from inference."),
+        ("Contradiction mapper", f"Read {context_file} and {findings_file}. Search for evidence that contradicts the current assumptions."),
+        ("Frontier mapper", f"Read {context_file}. List unknowns whose answers would change the next archetype."),
+    ], state_dir=state_dir, max_workers=4)
+
+with phase("Reduce map", "knowns, unknowns, contradictions, next archetype"):
+    frontier = parallel([
+        ("Exploration reducer", f"Read {findings_file} and outputs: {maps}. Write knowns, unknowns, contradictions, and recommended next archetype.")
+    ])[0]
+
+print(frontier)
+```
+
+### Sweep template
+
+```python
+from assets.ga_ultraplan import phase, mapchain, parallel
+
+state_dir = "./temp/ultraplan_<target>"
+context_file = f"{state_dir}/context.md"       # target-specific context or source references
+criteria_file = f"{state_dir}/criteria.md"     # optional constraints/checklist
+items_file = f"{state_dir}/items.txt"          # one item per line, produced or checked by the controller
+items = [line.strip() for line in open(items_file, encoding="utf-8") if line.strip()]
+
+with phase("Per-item sweep", "no duplicate or omitted items"):
+    item_reports = mapchain(
+        items,
+        ("Inspect {item}", f"Read {context_file} and {criteria_file}. Inspect assigned item only: {{item}}. Output item ID, evidence, proposed action, unresolved risk."),
+        ("Verify {previous}", "Tail {previous}. Verify the item report against tools/files. Keep only supported claims and note gaps."),
+        state_dir=state_dir,
+        max_workers=6,
+    )
+
+with phase("Coverage reduce", "coverage table and omissions"):
+    coverage = parallel([
+        ("Coverage reducer", f"Read item list {items_file} and reports: {item_reports}. Produce coverage count, omissions, accepted findings, and next missing-set sweep if needed.")
+    ])[0]
+
+print(coverage)
+```
+
+If you wrote `reviews = parallel(review tasks); verified = parallel(verify tasks)` but each item only needs its own prior review, rewrite it as `mapchain`.
+
+## Dynamic fan-out and parsing discipline
+
+Avoid asking workers to return complex JSON task graphs. For dynamic fan-out:
+
+1. Prefer controller scouting to produce a plain item/control list before the script.
+2. Or have a planner/reducer write a plain text list file with one item per line.
+3. Validate the list with Python before using it for `mapchain`.
+4. If the list changes the workflow shape, stop and regenerate the next script instead of contorting the current script.
+
+Only parse small, explicit control data. Rich reasoning belongs in `.out.txt` files and shared markdown files.
+
+## Output discipline
+
+Worker outputs should end with a compact, easy-to-tail conclusion:
+
+```text
+## Conclusion
+- Item / lens:
+- Evidence inspected:
+- Finding or attempt result:
+- Confidence / failure reason:
+- What should happen next:
+```
+
+Reducers should produce controller-ready files:
+
+```text
+## Decision state
+- Accepted facts:
+- Rejected / unproven claims:
+- Attempts or decisions recorded:
+- Remaining frontier:
+- Recommended next archetype:
+- Coverage bounds / skipped scope:
+```
+
+Downstream prompts should usually say `Tail <previous>.out.txt` rather than `Read <previous>.out.txt` to avoid re-ingesting the full transcript noise.
+
+## Final user-facing answer requirements
+
+When UltraPlan is used to produce a plan, the final answer must include:
+
+- what archetype(s) were run and what scope was covered;
+- the concrete decision or plan, not just that subagents completed;
+- ordered steps with responsible files/items when implementation is involved;
+- evidence or output files that support the key claims;
+- verification steps that can actually be executed;
+- risks, skipped scope, unresolved questions, and why they do or do not block progress;
+- whether another UltraPlan wave is recommended.
+
+Do not claim exhaustive coverage unless the item list, coverage notes, and reducer prove it. Do not pad with generic advice.
+
+## Safety and execution rules
+
+- Read relevant code, files, pages, logs, or records before planning edits. UltraPlan must be grounded in the actual target.
+- Keep irreversible operations out of worker prompts unless the user explicitly approved them.
+- Subagents have full tools and physical execution ability; give precise boundaries, allowed targets, and stop conditions.
+- Prefer file paths over copied long text in prompts. Tell workers exactly which shared files and output files to inspect.
+- After a script completes, the main agent must read the final output file and sanity-check important intermediate files.
+- Use several smaller scripts when the next phase depends on prior results; do not hide recursive control inside one huge script.
+- If a worker fails, inspect its `.out.txt` or error, update state, reduce scope or timeout, and retry with new information. Do not repeat blindly.
+- If coverage is bounded by time, sampling, top-N, timeout, excluded subsystem, or failed tool, record the bound in the relevant shared note and the final answer.
+
+## Minimal controller skeleton
+
+```python
+from assets.ga_ultraplan import phase, parallel, mapchain
+
+state_dir = "./temp/ultraplan_<target>"
+context_file = f"{state_dir}/context.md"          # task-shaped shared context, not a required schema
+attempts_file = f"{state_dir}/attempts.md"        # optional prior attempts or coverage notes
+criteria_file = f"{state_dir}/criteria.md"        # optional success criteria or constraints
+
+with phase("Current archetype", "state what this script is trying to learn or reduce"):
+    outputs = parallel([
+        ("Lens A", f"Read {context_file}, plus {attempts_file} or {criteria_file} if relevant. Do the assigned independent work."),
+        ("Lens B", f"Read {context_file}, plus {attempts_file} or {criteria_file} if relevant. Use a different lens; avoid duplicating Lens A."),
+    ], state_dir=state_dir)
+
+with phase("Reduce", "controller-ready decision state"):
+    reduced = parallel([
+        ("Reducer", f"Read the relevant shared notes and these outputs: {outputs}. Write accepted facts, rejected claims, frontier, and recommended next archetype.")
+    ])[0]
+
+print(reduced)
+```
+
+After this script exits, the main agent reads `reduced`, updates shared state, and decides whether to answer, act directly, ask the user, or write the next UltraPlan script.