drydock-cli 2.6.134__tar.gz → 2.6.136__tar.gz

{drydock_cli-2.6.134 → drydock_cli-2.6.136}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 2.6.134
+Version: 2.6.136
 Summary: Local-first CLI coding agent — tested with Gemma 4 26B via vLLM
 Project-URL: Homepage, https://github.com/fbobe321/drydock
 Project-URL: Repository, https://github.com/fbobe321/drydock

{drydock_cli-2.6.134 → drydock_cli-2.6.136}/drydock/core/agent_loop.py

@@ -643,11 +643,13 @@ class AgentLoop:
             tool_turns = 0
             api_error_count = 0
             has_made_edit = False  # Track if model has used search_replace/write_file
-            # Per-user-prompt wall-clock budget. Gemma 4 routinely
-            # spends 60+ minutes on a single prompt without closing it
-            # — looks like work is happening, but the user is staring
-            # at a non-responsive TUI. Cap at 15 minutes per prompt.
-            PER_PROMPT_BUDGET_SEC = 15 * 60
+            # Per-user-prompt wall-clock budget. Gemma 4 can spend 60+
+            # minutes on a single prompt without closing it, but user
+            # feedback (issue #9) showed 15 min was cutting off legitimate
+            # "really difficult" builds. Bumped to 30 min — long enough
+            # for a multi-file refactor, short enough that a runaway loop
+            # still hands control back before the user gives up.
+            PER_PROMPT_BUDGET_SEC = 30 * 60
             _prompt_start = time.perf_counter()
             logger.warning("[TIMING] entering conversation while loop")
             while not should_break_loop:
@@ -681,27 +683,26 @@ class AgentLoop:
                         stopped_by_middleware=True,
                     )
                     return
-                if tool_turns == 15:
+                if tool_turns == 30:
                     self._inject_system_note(
                         f"You have used {tool_turns} tool calls on this "
                         "single user request. Start wrapping up — make "
                         "your next 3-5 calls count, then stop with a "
                         "summary so the user can review."
                     )
-                elif tool_turns == 25:
+                elif tool_turns == 60:
                     self._inject_system_note(
                         f"You have used {tool_turns} tool calls on this "
                         "single request. STOP NOW. Emit a final text "
                         "response summarizing what you did (or what is "
                         "blocked) so the user can take the next step."
                     )
-                elif tool_turns >= 50:
+                elif tool_turns >= 100:
                     # Hard end-of-turn: synthesize a user-facing message
-                    # and stop. Was 35 but that truncated initial builds
-                    # (tool_agent stress: agent.py never got written,
-                    # breaking the whole subsequent run). 50 is generous
-                    # enough for a full build while still preventing
-                    # infinite loops.
+                    # and stop. Was 50 but issue #9 showed "really
+                    # difficult" tasks legitimately need more than 50
+                    # tool calls. 100 preserves the runaway-loop safety
+                    # while giving complex builds room to finish.
                     yield AssistantEvent(
                         content=(
                             f"\n\n[Drydock: stopped after {tool_turns} tool "

{drydock_cli-2.6.134 → drydock_cli-2.6.136}/drydock/core/prompts/gemma4.md

@@ -69,9 +69,18 @@ Rules:
   inline — put each numbered item on its own line.
 - NEVER ask "would you like me to proceed" or "shall I continue" — JUST DO IT.
 - NEVER stop to report progress or ask for confirmation between steps.
-  If you have a todo list with multiple items, execute ALL of them
-  without pausing. Only stop when EVERY item is done.
-- After creating/editing a file, move to the next one. Do not stop.
+- KNOW WHEN TO STOP. Two modes, and you must not confuse them:
+  * TODO MODE: if you created a todo list via the `todo` tool, or the
+    user gave a multi-step request ("build the package", "plan then
+    build", "fix all the failing tests"), execute ALL items without
+    pausing. Only stop when every item is done.
+  * SIMPLE MODE: if the user gave a single request ("fix this bug",
+    "add this method", "explain how X works"), do exactly that and
+    stop. Do NOT invent follow-up work ("let me also add a test",
+    "let me refactor this too") the user did not ask for. One ask,
+    one answer, then silence until the next prompt.
+  If you are about to do something the user did not explicitly ask
+  for, stop and let the user drive the next step.
 - Follow the EXACT CLI interface specified in the PRD. Match argument names, subcommands, and flags exactly.
 - Every subcommand in the PRD must have a working handler — not just argparse registration.
 - NEVER write `class X: def method(self): pass` inline in cli.py or __main__.py to silence ModuleNotFoundError. Write the REAL class in its own file (e.g. `interpreter.py`) and import it. Stub classes pass `--help` but fail every real task.

{drydock_cli-2.6.134 → drydock_cli-2.6.136}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "drydock-cli"
-version = "2.6.134"
+version = "2.6.136"
 description = "Local-first CLI coding agent — tested with Gemma 4 26B via vLLM"
 readme = "README.md"
 requires-python = ">=3.12"