ticket #76 attempt 1

devplacepy/cli/__init__.py

@@ -37,9 +37,10 @@ from devplacepy.cli.backups import (
 )
 from devplacepy.cli.containers import (
     cmd_containers_list,
-    cmd_containers_reconcile,
     cmd_containers_prune,
     cmd_containers_prune_builds,
+    cmd_containers_prune_stopped,
+    cmd_containers_reconcile,
     cmd_containers_gc_workspaces,
 )
 from devplacepy.cli.migrate import cmd_emoji_sync, cmd_migrate_data
@@ -83,6 +84,7 @@ __all__ = [
     "cmd_containers_reconcile",
     "cmd_containers_prune",
     "cmd_containers_prune_builds",
+    "cmd_containers_prune_stopped",
     "cmd_containers_gc_workspaces",
     "cmd_emoji_sync",
     "cmd_migrate_data",

devplacepy/cli/containers.py

@@ -1,5 +1,6 @@
 # retoor <retoor@molodetz.nl>
 
+import asyncio
 from devplacepy.cli._shared import _audit_cli
 
 
@@ -18,7 +19,6 @@ def cmd_containers_list(args):
 
 
 def cmd_containers_reconcile(args):
-    import asyncio
     from devplacepy.services.containers.service import ContainerService
 
     asyncio.run(ContainerService().run_once())
@@ -27,7 +27,6 @@ def cmd_containers_reconcile(args):
 
 
 def cmd_containers_prune(args):
-    import asyncio
     from devplacepy.services.containers.runtime import get_backend
     from devplacepy.services.containers.service import ContainerService
 
@@ -66,6 +65,29 @@ def cmd_containers_prune_builds(args):
     )
 
 
+def cmd_containers_prune_stopped(args):
+    from devplacepy.services.containers import api, store
+    from devplacepy.services.containers.service import ContainerService
+
+    async def run():
+        instances = store.all_instances()
+        marked = 0
+        for inst in instances:
+            if inst["status"] == store.ST_STOPPED and inst["desired_state"] == store.DESIRED_STOPPED:
+                api.mark_for_removal(inst, actor=("system", "prune-stopped"))
+                marked += 1
+        if marked:
+            await ContainerService().run_once()
+        return marked
+
+    marked = asyncio.run(run())
+    _audit_cli("cli.containers.prune_stopped", f"CLI removed {marked} stale stopped container(s)", metadata={"count": marked})
+    if marked:
+        print(f"Prune stopped: marked {marked} container(s) for removal and reconciled.")
+    else:
+        print("No stopped containers to prune.")
+
+
 def cmd_containers_gc_workspaces(args):
     import shutil
     from pathlib import Path
@@ -102,6 +124,9 @@ def register_containers(subparsers):
         "prune-builds",
         help="Remove legacy per-project images and clear the dockerfiles/builds tables",
     ).set_defaults(func=cmd_containers_prune_builds)
+    containers_sub.add_parser(
+        "prune-stopped", help="Remove all stopped containers (completed purpose)"
+    ).set_defaults(func=cmd_containers_prune_stopped)
     containers_sub.add_parser(
         "gc-workspaces", help="Remove workspace dirs with no instances"
     ).set_defaults(func=cmd_containers_gc_workspaces)

devplacepy/routers/tools/deepsearch.py

@@ -227,8 +227,7 @@ def _session_context(request: Request, uid: str, job: dict, session: dict) -> di
     user = get_current_user(request)
     viewer_is_admin = is_admin(user)
     done = bool(report) or job.get("status") == queue.DONE
-    cost_usd = report.get("cost_usd", 0.0) if report else 0.0
+    return {
-    ctx = {
         "uid": uid,
         "status": queue.DONE if done else job.get("status", ""),
         "query": report.get("query") or session.get("query"),
@@ -248,7 +247,6 @@ def _session_context(request: Request, uid: str, job: dict, session: dict) -> di
         "export_md_url": f"/tools/deepsearch/{uid}/export.md" if done else None,
         "export_json_url": f"/tools/deepsearch/{uid}/export.json" if done else None,
         "export_pdf_url": f"/tools/deepsearch/{uid}/export.pdf" if done else None,
-        "cost_usd": cost_usd if viewer_is_admin else None,
         "viewer_is_admin": viewer_is_admin,
         "viewer_owns": _owns(request, session),
         "created_at": session.get("created_at"),
@@ -257,7 +255,6 @@ def _session_context(request: Request, uid: str, job: dict, session: dict) -> di
         "user": user,
         "meta_robots": "noindex,nofollow",
     }
-    return ctx
 
 @router.get("/{uid}/session")
 async def deepsearch_session(request: Request, uid: str):

devplacepy/schemas/jobs.py

@@ -136,7 +136,6 @@ class DeepsearchSessionOut(_Out):
     export_md_url: Optional[str] = None
     export_json_url: Optional[str] = None
     export_pdf_url: Optional[str] = None
-    cost_usd: Optional[float] = None
     viewer_is_admin: bool = False
     viewer_owns: bool = False
     created_at: Optional[str] = None

devplacepy/services/deepsearch/llm.py

@@ -7,7 +7,6 @@ import time
 
 from devplacepy import stealth
 from devplacepy.config import INTERNAL_GATEWAY_URL, INTERNAL_MODEL
-from devplacepy.services.openai_gateway.usage import parse_usage_headers
 
 logger = logging.getLogger(__name__)
 
@@ -42,10 +41,7 @@ async def request_completion(
     if response.status_code >= 400:
         raise RuntimeError(f"chat gateway returned {response.status_code}")
     data: dict = response.json()
-    cost_info = parse_usage_headers(response.headers) or {}
+    return data, data.get("usage") or {}, elapsed_ms
-    usage = data.get("usage") or {}
-    usage["cost_usd"] = cost_info.get("cost_usd", 0.0)
-    return data, usage, elapsed_ms
 
 
 async def complete_chat(

devplacepy/services/jobs/deepsearch/enhance.py

@@ -5,11 +5,10 @@ from __future__ import annotations
 import json
 import logging
 import re
-from typing import Callable, Tuple
+from typing import Callable
 
 from devplacepy import stealth
 from devplacepy.config import INTERNAL_GATEWAY_URL, INTERNAL_MODEL
-from devplacepy.services.openai_gateway.usage import parse_usage_headers
 
 from .phases import PHASE_PLANNING
 
@@ -62,7 +61,7 @@ def _noop(frame: dict) -> None:
 
 async def plan_queries(
     query: str, api_key: str, emit: Callable[[dict], None] = _noop
-) -> Tuple[list[str], float]:
+) -> list[str]:
     emit(
         {
             "type": "substep",
@@ -90,8 +89,6 @@ async def plan_queries(
             )
         if response.status_code >= 400:
             raise RuntimeError(f"planner gateway returned {response.status_code}")
-        cost_info = parse_usage_headers(response.headers) or {}
-        planner_cost = cost_info.get("cost_usd", 0.0)
         data = response.json()
         content = (
             (data.get("choices") or [{}])[0].get("message", {}).get("content") or ""
@@ -105,7 +102,7 @@ async def plan_queries(
                     "message": f"Planned {len(parsed)} search angles",
                 }
             )
-            return parsed, planner_cost
+            return parsed
     except Exception as exc:
         logger.warning("deepsearch query planner failed, using fallback: %s", exc)
     fallback = _fallback(query)
@@ -116,4 +113,4 @@ async def plan_queries(
             "message": f"Planner unavailable, using {len(fallback)} heuristic angles",
         }
     )
-    return fallback, 0.0
+    return fallback

devplacepy/services/jobs/deepsearch/orchestrate.py

@@ -40,7 +40,6 @@ class Orchestration:
     source_diversity: float = 0.0
     score: int = 0
     synthesis: str = "agents"
-    total_cost_usd: float = 0.0
 
 
 def _domain(url: str) -> str:
@@ -166,7 +165,6 @@ async def _complete(
         "tokens_in": int(raw_usage.get("prompt_tokens") or prompt_chars // 4),
         "tokens_out": int(raw_usage.get("completion_tokens") or len(text) // 4),
         "elapsed_ms": elapsed_ms,
-        "cost_usd": float(raw_usage.get("cost_usd", 0.0)),
     }
     return text, usage
 
@@ -298,7 +296,6 @@ def _agent_done(emit: Callable[[dict], None], agent: str, usage: dict) -> None:
             "elapsed_ms": usage.get("elapsed_ms", 0),
             "tokens_in": usage.get("tokens_in", 0),
             "tokens_out": usage.get("tokens_out", 0),
-            "cost_usd": usage.get("cost_usd", 0.0),
         }
     )
 
@@ -349,7 +346,6 @@ async def orchestrate(
     emit: Callable[[dict], None],
     store=None,
     queries: list[str] | None = None,
-    planner_cost_usd: float = 0.0,
 ) -> Orchestration:
     diversity = source_diversity(pages)
     if not pages:
@@ -422,12 +418,6 @@ async def orchestrate(
         score = int(
             min(SCORE_MAX, (confidence * 0.5 + diversity * 0.3 + coverage * 0.2) * SCORE_MAX)
         )
-        total_cost = (
-            planner_cost_usd
-            + float(summary_usage.get("cost_usd", 0.0))
-            + float(findings_usage.get("cost_usd", 0.0))
-            + float(linker_usage.get("cost_usd", 0.0))
-        )
         return Orchestration(
             summary=summary,
             findings=findings,
@@ -435,7 +425,6 @@ async def orchestrate(
             source_diversity=diversity,
             score=score,
             synthesis="agents",
-            total_cost_usd=total_cost,
         )
     except Exception as exc:
         logger.warning("deepsearch orchestration failed, using heuristic: %s", exc)

devplacepy/services/jobs/deepsearch/worker.py

@@ -165,7 +165,7 @@ async def _run(payload: dict, output_dir: Path) -> dict:
     should_stop = _make_stop(output_dir)
 
     _stage("planning", "Planning research queries", PHASE_PLANNING)
-    queries, planner_cost = await plan_queries(query, api_key, _emit)
+    queries = await plan_queries(query, api_key, _emit)
     _emit({"type": "queries", "queries": queries})
 
     _stage("searching", "Searching the web", PHASE_SEARCHING)
@@ -203,8 +203,7 @@ async def _run(payload: dict, output_dir: Path) -> dict:
 
     _stage("analysis", "Running research agents", PHASE_ANALYSIS)
     result = await orchestrate(
-        query, outcome.pages, api_key, _emit, store=store, queries=queries,
+        query, outcome.pages, api_key, _emit, store=store, queries=queries
-        planner_cost_usd=planner_cost,
     )
 
     _stage("synthesis", "Compiling cited report", PHASE_SYNTHESIS)
@@ -229,7 +228,6 @@ async def _run(payload: dict, output_dir: Path) -> dict:
         "chunk_count": chunk_count,
         "embed_backend": embed_backend,
         "collection": collection,
-        "cost_usd": result.total_cost_usd,
     }
     (output_dir / "report.json").write_text(
         json.dumps(report, ensure_ascii=False), encoding="utf-8"
@@ -246,7 +244,6 @@ async def _run(payload: dict, output_dir: Path) -> dict:
             "synthesis": report["synthesis"],
             "page_count": report["page_count"],
             "chunk_count": report["chunk_count"],
-            "cost_usd": report["cost_usd"],
         }
     )
     return report

devplacepy/templates/docs/services-containers.html

@@ -107,7 +107,7 @@ instead of the gateway. Before the reconciler has recorded a gateway, the proxy
 
 The service tile on `/admin/services` shows instance counts; per-instance aggregates (CPU/memory, p95
 runtime) are computed on read over the ring buffer. The CLI exposes
-`devplace containers list | reconcile | prune | prune-builds | gc-workspaces` (`prune-builds` is a
+`devplace containers list | reconcile | prune | prune-builds | prune-stopped | gc-workspaces` (`prune-builds` is a
 one-time cleanup that removes legacy per-project images and the old dockerfiles/builds tables). Devii's
 admin `container_*` tools cover the same instance operations in natural language.
 </div>

devplacepy/templates/tools/deepsearch_session.html

@@ -16,9 +16,6 @@
                 <span class="ds-metric"><strong>{{ source_diversity }}</strong> diversity</span>
                 <span class="ds-metric"><strong>{{ page_count }}</strong> sources</span>
                 <span class="ds-metric"><strong>{{ chunk_count }}</strong> chunks</span>
-                {% if viewer_is_admin and cost_usd is not none %}
-                <span class="ds-metric"><strong>${{ "%.6f"|format(cost_usd) }}</strong> cost</span>
-                {% endif %}
             </div>
             {% if export_md_url %}
             <div class="ds-exports">

dpc.log

@@ -1,8 +1,24 @@
-2026-07-19T17:31:36 INFO logging initialised at /workspace/repo/dpc.log
+2026-07-19T09:32:07 INFO logging initialised at /workspace/repo/dpc.log
-2026-07-19T17:31:36 DEBUG model=molodetz-pro fps=30
+2026-07-19T09:32:07 DEBUG model=molodetz-pro fps=30
-2026-07-19T17:31:36 INFO read task from file: /workspace/prompts/research-10.txt
+2026-07-19T09:32:07 INFO read task from file: /workspace/prompts/research-1.txt
-2026-07-19T17:31:36 INFO settings merged: model=<default> allow=0 deny=0 ask=0
+2026-07-19T09:32:07 INFO settings merged: model=<default> allow=0 deny=0 ask=0
-2026-07-19T19:05:24 INFO logging initialised at /workspace/repo/dpc.log
+2026-07-19T15:54:39 INFO logging initialised at /workspace/repo/dpc.log
-2026-07-19T19:05:24 DEBUG model=molodetz-pro fps=30
+2026-07-19T15:54:39 DEBUG model=molodetz-pro fps=30
-2026-07-19T19:05:24 INFO read task from file: /workspace/prompts/execution-1.txt
+2026-07-19T15:54:39 INFO read task from file: /workspace/prompts/research-2.txt
-2026-07-19T19:05:24 INFO settings merged: model=<default> allow=0 deny=0 ask=0
+2026-07-19T15:54:39 INFO settings merged: model=<default> allow=0 deny=0 ask=0
+2026-07-19T17:25:31 INFO logging initialised at /workspace/repo/dpc.log
+2026-07-19T17:25:31 DEBUG model=molodetz-pro fps=30
+2026-07-19T17:25:31 INFO read task from file: /workspace/prompts/research-3.txt
+2026-07-19T17:25:31 INFO settings merged: model=<default> allow=0 deny=0 ask=0
+2026-07-19T17:37:32 INFO logging initialised at /workspace/repo/dpc.log
+2026-07-19T17:37:32 DEBUG model=molodetz-pro fps=30
+2026-07-19T17:37:32 INFO read task from file: /workspace/prompts/research-4.txt
+2026-07-19T17:37:32 INFO settings merged: model=<default> allow=0 deny=0 ask=0
+2026-07-19T18:22:18 INFO logging initialised at /workspace/repo/dpc.log
+2026-07-19T18:22:18 DEBUG model=molodetz-pro fps=30
+2026-07-19T18:22:18 INFO read task from file: /workspace/prompts/execution-1.txt
+2026-07-19T18:22:18 INFO settings merged: model=<default> allow=0 deny=0 ask=0
+2026-07-19T18:49:24 INFO logging initialised at /workspace/repo/dpc.log
+2026-07-19T18:49:24 DEBUG model=molodetz-pro fps=30
+2026-07-19T18:49:24 INFO read task from file: /workspace/prompts/execution-2.txt
+2026-07-19T18:49:24 INFO settings merged: model=<default> allow=0 deny=0 ask=0

tests/api/tools/deepsearch/session.py

@@ -39,7 +39,6 @@ def _seed_done_session(owner_id="ds-session-owner"):
         "source_diversity": 0.5,
         "page_count": 3,
         "chunk_count": 12,
-        "cost_usd": 0.123456,
     }
     result = {
         "query": "the question",
@@ -48,7 +47,6 @@ def _seed_done_session(owner_id="ds-session-owner"):
         "source_diversity": 0.5,
         "page_count": 3,
         "chunk_count": 12,
-        "cost_usd": 0.123456,
         "report": report,
     }
     get_table("jobs").update(
@@ -99,7 +97,6 @@ def test_session_json_shape(app_server):
         assert body["sources"]
         assert "viewer_is_admin" in body
         assert "viewer_owns" in body
-        assert body.get("cost_usd") is None  # guest user, not admin
     finally:
         _clear()
 
@@ -111,7 +108,6 @@ def test_session_html_renders_without_jinja_global_collision(app_server):
         assert r.status_code == 200, r.text
         assert "the question" in r.text
         assert "dp-deepsearch-chat" in r.text
-        assert "0.123456" not in r.text  # guest user, cost hidden
     finally:
         _clear()
 
@@ -146,7 +142,6 @@ def test_session_reads_disk_report_before_result_commit(app_server):
         "synthesis": "agents",
         "page_count": 12,
         "chunk_count": 61,
-        "cost_usd": 0.123456,
     }
     session_dir = DEEPSEARCH_DIR / uid
     session_dir.mkdir(parents=True, exist_ok=True)
@@ -208,26 +203,3 @@ def test_export_json(app_server):
 def test_export_unknown_uid_404(app_server):
     r = requests.get(f"{BASE_URL}/tools/deepsearch/nope/export.md")
     assert r.status_code == 404
-
-
-def test_session_admin_sees_cost(app_server, seeded_db):
-    try:
-        uid = _seed_done_session("alice_test")
-        s = requests.Session()
-        creds = seeded_db["alice"]
-        s.post(
-            f"{BASE_URL}/auth/login",
-            data={"email": creds["email"], "password": creds["password"]},
-            allow_redirects=True,
-        )
-        r = s.get(f"{BASE_URL}/tools/deepsearch/{uid}/session", headers=_json_headers())
-        assert r.status_code == 200, r.text
-        body = r.json()
-        assert body.get("cost_usd") == 0.123456
-
-        html = s.get(f"{BASE_URL}/tools/deepsearch/{uid}/session")
-        assert html.status_code == 200, html.text
-        assert "0.123456" in html.text
-        assert "cost" in html.text
-    finally:
-        _clear()