drydock-cli 3.0.76__tar.gz → 3.0.77__tar.gz

{drydock_cli-3.0.76/drydock_cli.egg-info → drydock_cli-3.0.77}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 3.0.76
+Version: 3.0.77
 Summary: Drydock — a local, provider-agnostic terminal coding agent for local LLMs
 Author: Frank Bobe III
 License-Expression: Apache-2.0
@@ -59,6 +59,10 @@ A full agentic CLI harness — every tool below is clean-room and dependency-fre
 
 - **Files & shell** — `Read` (with a structure index for huge files), `Write`,
   `Edit`, `Bash`, `Glob`, `Grep`.
+- **Vision (multimodal)** — reference an image path in your message (a `.png`/
+  `.jpg` screenshot, mockup, or diagram) and it's attached for a vision-capable
+  model to *see* — describe a UI, read text off a screenshot, debug a diagram.
+  Works with any `--mmproj`-enabled server (e.g. llama.cpp with a vision model).
 - **Version control** — `GitStatus`, `GitDiff`, `GitLog`, `GitCommit`
   (structured + truncated; commit is local and reversible).
 - **Internet** — `WebSearch` + `WebFetch` (DuckDuckGo; offline-safe).

{drydock_cli-3.0.76 → drydock_cli-3.0.77}/README.md

@@ -35,6 +35,10 @@ A full agentic CLI harness — every tool below is clean-room and dependency-fre
 
 - **Files & shell** — `Read` (with a structure index for huge files), `Write`,
   `Edit`, `Bash`, `Glob`, `Grep`.
+- **Vision (multimodal)** — reference an image path in your message (a `.png`/
+  `.jpg` screenshot, mockup, or diagram) and it's attached for a vision-capable
+  model to *see* — describe a UI, read text off a screenshot, debug a diagram.
+  Works with any `--mmproj`-enabled server (e.g. llama.cpp with a vision model).
 - **Version control** — `GitStatus`, `GitDiff`, `GitLog`, `GitCommit`
   (structured + truncated; commit is local and reversible).
 - **Internet** — `WebSearch` + `WebFetch` (DuckDuckGo; offline-safe).

{drydock_cli-3.0.76 → drydock_cli-3.0.77}/drydock/providers.py

@@ -250,16 +250,11 @@ _IMAGE_MIME = {
 }
 
 
-def _user_content_with_images(content):
-    """Vision support: if the user's text references image file paths that exist
-    on disk, attach them as OpenAI multimodal image_url blocks (works with any
-    --mmproj-enabled server). Text-only prompts pass through unchanged as a plain
-    string, so display / loop-detection / compaction / token-counting (which all
-    assume string content) are untouched — the multimodal list is built ONLY here,
-    at the API boundary."""
+def detect_image_paths(content) -> list[str]:
+    """Image file paths referenced in a user message that exist on disk. Shared
+    by the multimodal attach (below) and the TUI (to confirm '📎 attached')."""
     if not isinstance(content, str) or "." not in content:
-        return content
-    import base64
+        return []
     import os
     import re
 
@@ -281,8 +276,22 @@ def _user_content_with_images(content):
         p = os.path.expanduser(raw)
         if os.path.isfile(p) and p not in seen:
             seen.append(p)
+    return seen
+
+
+def _user_content_with_images(content):
+    """Vision support: if the user's text references image file paths that exist
+    on disk, attach them as OpenAI multimodal image_url blocks (works with any
+    --mmproj-enabled server). Text-only prompts pass through unchanged as a plain
+    string, so display / loop-detection / compaction / token-counting (which all
+    assume string content) are untouched — the multimodal list is built ONLY here,
+    at the API boundary."""
+    seen = detect_image_paths(content)
     if not seen:
         return content
+    import base64
+    import os
+
     blocks: list[dict] = [{"type": "text", "text": content}]
     for p in seen:
         mime = _IMAGE_MIME.get(os.path.splitext(p)[1].lower(), "image/png")

{drydock_cli-3.0.76 → drydock_cli-3.0.77}/drydock/tui/app.py

@@ -409,6 +409,14 @@ class DrydockApp(App):
         # running, queue this one and drain it when the current turn finishes
         # instead of dropping it on the floor.
         self._mount(UserMessage(text))
+        # Confirm any image attachments so the user SEES vision is active (the
+        # actual attach happens at the API boundary in providers).
+        from drydock import providers
+        imgs = providers.detect_image_paths(text)
+        if imgs:
+            import os as _os
+            names = ", ".join(_os.path.basename(p) for p in imgs)
+            self._info(f"📎 attached {len(imgs)} image(s) for the model to see: {names}")
         if self._busy:
             self._queue.append(text)
             self._refresh_status()

{drydock_cli-3.0.76 → drydock_cli-3.0.77}/drydock/tuning.py

@@ -90,7 +90,11 @@ _DEFAULT_SYSTEM_PROMPT = (
     "start to lay out the plan, then START WORKING THROUGH IT IMMEDIATELY — do "
     "NOT stop to ask whether to proceed. Call `todo` again to flip a step to "
     "done and the next to in-progress as you finish each, and only stop when "
-    "every step is done. Don't write a TODO.md file for this."
+    "every step is done. Don't write a TODO.md file for this.\n"
+    "You have VISION: when the user references an image file path (.png/.jpg/"
+    ".jpeg/.gif/.webp/.bmp on disk — a screenshot, mockup, diagram, or photo), it "
+    "is attached and you can SEE it; describe it, read text from it, or debug "
+    "with it — never claim you cannot view images."
 )
 
 # Short, imperative prompt. Small local models do better with "act now" than
@@ -112,7 +116,11 @@ _GEMMA_SYSTEM_PROMPT = (
     "(one task per line, '[ ]'/'[~]'/'[x]'), then IMMEDIATELY start doing the "
     "steps — do NOT stop to ask whether to proceed. Call `todo` again to update "
     "it as you finish each step, and only stop when every step is done. Do not "
-    "write a TODO.md file for this."
+    "write a TODO.md file for this.\n"
+    "You have VISION: when the user's message references an image file path "
+    "(.png/.jpg/.jpeg/.gif/.webp/.bmp that exists on disk — a screenshot, mockup, "
+    "diagram, or photo), that image is attached and you can SEE it. Describe it, "
+    "read text from it, or use it to debug — never claim you cannot view images."
 )
 
 
@@ -221,11 +229,15 @@ _DRYDOCK_COMMANDS_HELP = (
     "- Custom skills (reusable `/<name>` prompts): `/skills new <name> <prompt "
     "text>` creates one (use $ARGS in the prompt for trailing input); `/skills` "
     "lists them; then they run it as `/<name>`.\n"
+    "- Images / vision: the user just references an image file path in their "
+    "message (a .png/.jpg screenshot, mockup, or diagram on disk) and it is "
+    "attached for you to SEE — no command needed.\n"
     "- Other: `/model` (model/endpoint), `/cwd`, `/undo` (revert last write), "
-    "`/back` (rewind a turn), `/compact` (shrink context), `/loop <n> <prompt>` "
-    "(repeat a prompt), `/mcp` (connected MCP servers), `/status`, `/clear`, "
-    "`/help`, `/quit`. Internet + git are tools you call yourself "
-    "(WebSearch/WebFetch, GitStatus/GitDiff/GitLog/GitCommit)."
+    "`/back` (rewind a turn), `/compact` (shrink context), `/context` (view/set "
+    "the context-window budget), `/loop <n> <prompt>` (repeat a prompt), `/mcp` "
+    "(connected MCP servers), `/rmf` & `/stig` (NIST 800-53 / DISA STIG), "
+    "`/status`, `/clear`, `/help`, `/quit`. Internet + git are tools you call "
+    "yourself (WebSearch/WebFetch, GitStatus/GitDiff/GitLog/GitCommit)."
 )
 
 

{drydock_cli-3.0.76 → drydock_cli-3.0.77/drydock_cli.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 3.0.76
+Version: 3.0.77
 Summary: Drydock — a local, provider-agnostic terminal coding agent for local LLMs
 Author: Frank Bobe III
 License-Expression: Apache-2.0
@@ -59,6 +59,10 @@ A full agentic CLI harness — every tool below is clean-room and dependency-fre
 
 - **Files & shell** — `Read` (with a structure index for huge files), `Write`,
   `Edit`, `Bash`, `Glob`, `Grep`.
+- **Vision (multimodal)** — reference an image path in your message (a `.png`/
+  `.jpg` screenshot, mockup, or diagram) and it's attached for a vision-capable
+  model to *see* — describe a UI, read text off a screenshot, debug a diagram.
+  Works with any `--mmproj`-enabled server (e.g. llama.cpp with a vision model).
 - **Version control** — `GitStatus`, `GitDiff`, `GitLog`, `GitCommit`
   (structured + truncated; commit is local and reversible).
 - **Internet** — `WebSearch` + `WebFetch` (DuckDuckGo; offline-safe).

{drydock_cli-3.0.76 → drydock_cli-3.0.77}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 # PyPI distribution name is drydock-cli (the established install name, continued
 # from the retired v2 fork); the import package + CLI command stay `drydock`.
 name = "drydock-cli"
-version = "3.0.76"
+version = "3.0.77"
 description = "Drydock — a local, provider-agnostic terminal coding agent for local LLMs"
 readme = "README.md"
 requires-python = ">=3.11"

{drydock_cli-3.0.76 → drydock_cli-3.0.77}/tests/test_vision_input.py

@@ -82,3 +82,23 @@ def test_image_path_in_backticks_and_punctuation(tmp_path):
         out = _user_content_with_images(prompt)
         assert isinstance(out, list), f"not attached: {prompt!r}"
         assert any(b.get("type") == "image_url" for b in out), prompt
+
+
+def test_detect_image_paths_shared_helper(tmp_path):
+    from drydock.providers import detect_image_paths
+    img = _write_png(tmp_path / "shot.png")
+    # plain, quoted, backticked, and trailing-punctuation forms all resolve
+    assert detect_image_paths(f"look at {img}") == [img]
+    assert detect_image_paths(f'see "{img}".') == [img]
+    assert detect_image_paths(f"debug `{img}`,") == [img]
+    # nonexistent path or text-only → nothing attached
+    assert detect_image_paths(f"{tmp_path}/nope.png") == []
+    assert detect_image_paths("no images here, just prose") == []
+    # de-dupes repeats
+    assert detect_image_paths(f"{img} and again {img}") == [img]
+
+
+def test_vision_in_system_prompt():
+    from drydock.tuning import system_prompt_for_model
+    assert "VISION" in system_prompt_for_model("gemma4")
+    assert "VISION" in system_prompt_for_model("some-other-model")