feat(llm): stop sending images for the rest of the run once the model rejects one

The image-strip recovery fired on any 400/404/422, wiping a session's screenshots and replaying the turn on unrelated errors (e.g. an upstream incomplete-tool-call 400, a context overflow). Require an image-related message too.

strix/agents/factory.py

@@ -18,6 +18,7 @@ from pydantic import ValidationError
 
 from strix.agents.prompt import render_system_prompt
 from strix.config import load_settings
+from strix.config.models import images_allowed
 from strix.tools.agents_graph.tools import (
     agent_finish,
     create_agent,
@@ -375,11 +376,16 @@ def _configure_filesystem_tools(
             )
 
 
+def _view_image_enabled(ctx: Any, _agent: Any) -> bool:
+    return images_allowed(ctx.context)
+
+
 def _make_filesystem_configurator(*, chat_completions: bool, strict_schemas: bool) -> Any:
     def configure(toolset: Any) -> None:
         _configure_filesystem_tools(
             toolset, chat_completions=chat_completions, strict_schemas=strict_schemas
         )
+        toolset.view_image.is_enabled = _view_image_enabled
 
     return configure
 
@@ -670,6 +676,7 @@ def build_strix_agent(
     system_prompt_context: dict[str, Any] | None = None,
     extra_tools: Sequence[Tool] | None = None,
     instructions_override: str | None = None,
+    supports_images: bool = True,
 ) -> SandboxAgent[Any]:
     """Build a SandboxAgent for either root or child use.
 
@@ -694,6 +701,7 @@ def build_strix_agent(
             is_diff_scoped=is_diff_scoped,
             interactive=interactive,
             system_prompt_context=system_prompt_context,
+            supports_images=supports_images,
         )
 
     agent_tools = [*_EXTRA_TOOLS, *(extra_tools or [])]
@@ -754,6 +762,7 @@ def make_child_factory(
     chat_completions_tools: bool = False,
     strict_tool_schemas: bool = True,
     system_prompt_context: dict[str, Any] | None = None,
+    supports_images: bool = True,
 ) -> Any:
     """Return the runner-owned builder used by ``spawn_child_agent``.
 
@@ -774,6 +783,7 @@ def make_child_factory(
             chat_completions_tools=chat_completions_tools,
             strict_tool_schemas=strict_tool_schemas,
             system_prompt_context=system_prompt_context,
+            supports_images=supports_images,
         )
 
     return _factory

strix/agents/prompt.py

@@ -85,6 +85,7 @@ def render_system_prompt(
     interactive: bool = False,
     system_prompt_context: dict[str, Any] | None = None,
     include_scope: bool = True,
+    supports_images: bool = True,
 ) -> str:
     """Render the system prompt. Returns empty string on template failure.
 
@@ -119,7 +120,7 @@ def render_system_prompt(
             is_root=is_root,
             is_diff_scoped=is_diff_scoped,
         )
-        skill_content = load_skills(skills_to_load)
+        skill_content = load_skills(skills_to_load, supports_images=supports_images)
         env.globals["get_skill"] = lambda name: skill_content.get(name, "")
 
         # Skills every agent of this kind loads come first, so siblings share them

strix/config/models.py

@@ -7,6 +7,7 @@ import contextlib
 import inspect
 import logging
 import os
+import threading
 import time
 import uuid
 from collections.abc import AsyncGenerator
@@ -905,6 +906,28 @@ def model_supports_reasoning(model_name: str) -> bool:
     return bool(entry and entry.get("supports_reasoning"))
 
 
+def model_supports_images(model_name: str) -> bool:
+    """Return whether the model accepts image input. Assume yes until proven otherwise."""
+    entry = _catalog_entry(model_name)
+    return entry is None or bool(entry.get("supports_vision"))
+
+
+# Set once the model rejects an image; one process is one run, so this covers every agent.
+_IMAGES_REJECTED = threading.Event()
+
+
+def mark_images_rejected() -> None:
+    if not _IMAGES_REJECTED.is_set():
+        logger.warning("The model rejected image input; sending no images for the rest of the run")
+        _IMAGES_REJECTED.set()
+
+
+def images_allowed(context: object) -> bool:
+    """Whether a request may carry images: the model should take them and hasn't refused one."""
+    supported = context.get("supports_images", True) if isinstance(context, dict) else True
+    return bool(supported) and not _IMAGES_REJECTED.is_set()
+
+
 def _bare_openai_name(model_name: str) -> str:
     name = model_name.strip().lower()
     for prefix in ("litellm/", "any-llm/", "openai/"):

strix/core/execution.py

@@ -4,9 +4,12 @@ from __future__ import annotations
 
 import asyncio
 import contextlib
+import inspect
 import logging
+import re
 import uuid
 from collections.abc import Callable
+from dataclasses import replace
 from functools import cache
 from typing import TYPE_CHECKING, Any, cast
 
@@ -20,6 +23,7 @@ from openai import (
 )
 
 from strix.config import codex
+from strix.config.models import images_allowed, mark_images_rejected
 from strix.core.hooks import (
     BudgetExceededError,
     BudgetPausedError,
@@ -30,6 +34,7 @@ from strix.core.sessions import (
     enforce_image_budget,
     open_agent_session,
     replace_session_items,
+    scrub_images_from_items,
     seed_initial_input,
     strip_all_images_from_session,
 )
@@ -44,6 +49,7 @@ if TYPE_CHECKING:
     from agents.lifecycle import RunHooks
     from agents.memory import Session, SQLiteSession
     from agents.result import RunResultBase
+    from agents.run_config import CallModelData, ModelInputData
 
     from strix.core.agents import AgentCoordinator, Status
 
@@ -53,6 +59,16 @@ logger = logging.getLogger(__name__)
 StreamEventSink = Callable[[str, Any], None]
 
 _INPUT_REJECTION_CODES = frozenset({400, 404, 422})
+# Replies meaning "this model takes no images", not image errors in general: a
+# context overflow that counts "image/vision expansion" must not match.
+_IMAGE_REJECTION = re.compile(
+    r"no endpoints found that support image input"  # OpenRouter
+    r"|image_url is only supported by certain models"  # OpenAI
+    r"|is not a multimodal model|at most 0 image\(s\)"  # vLLM
+    r"|does not support image input"  # LiteLLM's Fireworks check
+    r"|doesn't support the image field",  # Bedrock Converse
+    re.IGNORECASE,
+)
 _MAX_COMPACTIONS_PER_CYCLE = 2
 
 
@@ -120,6 +136,28 @@ async def _compact_session(
     )
 
 
+_TEXT_ONLY_IMAGE_TEXT = "[error: this model cannot view images; use `snapshot -i` instead]"
+
+
+def _with_image_scrub(run_config: RunConfig, context: dict[str, Any]) -> RunConfig:
+    if images_allowed(context):
+        return run_config
+    # Chain any filter already set; it sees the scrubbed input.
+    inner = run_config.call_model_input_filter
+
+    async def _scrub(data: CallModelData[Any]) -> ModelInputData:
+        model_data = replace(
+            data.model_data,
+            input=scrub_images_from_items(data.model_data.input, text=_TEXT_ONLY_IMAGE_TEXT),
+        )
+        if inner is None:
+            return model_data
+        result = inner(replace(data, model_data=model_data))
+        return await result if inspect.isawaitable(result) else result
+
+    return replace(run_config, call_model_input_filter=_scrub)
+
+
 _MAX_TRANSIENT_MODEL_RETRIES = 5
 _TRANSIENT_MODEL_RETRY_BASE_DELAY_S = 2.0
 _TRANSIENT_MODEL_RETRY_MAX_DELAY_S = 90.0
@@ -130,6 +168,12 @@ def _model_error_status_code(exc: BaseException) -> int | None:
     return code if isinstance(code, int) else None
 
 
+def _is_image_rejection(exc: BaseException) -> bool:
+    return _model_error_status_code(exc) in _INPUT_REJECTION_CODES and bool(
+        _IMAGE_REJECTION.search(str(exc))
+    )
+
+
 def _is_transient_model_error(exc: BaseException) -> bool:
     if codex.is_content_guardrail_error(exc):
         return False
@@ -747,7 +791,7 @@ async def _run_cycle(  # noqa: PLR0912, PLR0915
             stream = Runner.run_streamed(
                 agent,
                 input=input_data,
-                run_config=run_config,
+                run_config=_with_image_scrub(run_config, context),
                 context=context,
                 max_turns=max_turns,
                 session=session,
@@ -805,11 +849,10 @@ async def _run_cycle(  # noqa: PLR0912, PLR0915
             await coordinator.trigger_budget_stop()
             raise
         except Exception as exc:
-            if (
-                image_strips < 3
-                and session is not None
-                and getattr(exc, "status_code", None) in _INPUT_REJECTION_CODES
-            ):
+            image_rejected = _is_image_rejection(exc)
+            if image_rejected:
+                mark_images_rejected()
+            if image_strips < 3 and session is not None and image_rejected:
                 try:
                     stripped = await strip_all_images_from_session(session)
                 except Exception:

strix/core/runner.py

@@ -23,6 +23,7 @@ from strix.config.models import (
     StrixProvider,
     configure_sdk_api_route,
     configure_sdk_model_defaults,
+    model_supports_images,
     supports_strict_tool_schemas,
     uses_chat_completions_tool_schema,
 )
@@ -153,6 +154,7 @@ def _compose_root_instructions_override(
     is_diff_scoped: bool,
     interactive: bool,
     system_prompt_context: dict[str, Any],
+    supports_images: bool,
 ) -> str | None:
     if root_instructions_override is None:
         return None
@@ -166,6 +168,7 @@ def _compose_root_instructions_override(
         interactive=interactive,
         system_prompt_context=system_prompt_context,
         include_scope=False,
+        supports_images=supports_images,
     )
     return (
         f"{base_instructions}\n\n"
@@ -265,6 +268,9 @@ async def run_strix_scan(
     strict_tool_schemas = supports_strict_tool_schemas(resolved_model)
     if not strict_tool_schemas:
         logger.info("Sending non-strict tool schemas: %s caps strict tools", resolved_model)
+    supports_images = model_supports_images(resolved_model)
+    if not supports_images:
+        logger.info("Leaving out image tools: %s does not accept images", resolved_model)
 
     if budget_policy not in ("stop", "pause"):
         raise ValueError(f"unknown budget_policy: {budget_policy!r}")
@@ -468,6 +474,7 @@ async def run_strix_scan(
             is_diff_scoped=is_diff_scoped,
             interactive=interactive,
             system_prompt_context=root_context,
+            supports_images=supports_images,
         )
 
         root_agent = build_strix_agent(
@@ -482,6 +489,7 @@ async def run_strix_scan(
             strict_tool_schemas=strict_tool_schemas,
             system_prompt_context=root_context,
             instructions_override=root_instructions,
+            supports_images=supports_images,
         )
 
         if not is_resume:
@@ -501,6 +509,7 @@ async def run_strix_scan(
             chat_completions_tools=chat_completions_tools,
             strict_tool_schemas=strict_tool_schemas,
             system_prompt_context=scope_context,
+            supports_images=supports_images,
         )
 
         async def spawn_child_agent(**kwargs: Any) -> dict[str, Any]:
@@ -528,6 +537,7 @@ async def run_strix_scan(
             "spawn_child_agent": spawn_child_agent,
             "scan_targets": build_scan_targets(scan_config),
             "max_context_images": settings.runtime.max_context_images,
+            "supports_images": supports_images,
         }
 
         root_session = open_agent_session(root_id, agents_db)

strix/core/sessions.py

@@ -199,13 +199,13 @@ async def enforce_image_budget(session: Session, max_images: int) -> bool:
     return await _rewrite_session(session, _transform)
 
 
-def scrub_images_from_items(items: list[Any]) -> list[Any]:
+def scrub_images_from_items(items: list[Any], *, text: str = _INHERITED_IMAGE_TEXT) -> list[Any]:
     """Return a copy of ``items`` with every image block replaced by text."""
 
     def _scrub(obj: Any) -> Any:
         if isinstance(obj, dict):
             if obj.get("type") == "input_image":
-                return {"type": "input_text", "text": _INHERITED_IMAGE_TEXT}
+                return {"type": "input_text", "text": text}
             return {k: _scrub(v) for k, v in obj.items()}
         if isinstance(obj, list):
             return [_scrub(v) for v in obj]

strix/llm/compaction.py

@@ -15,11 +15,7 @@ from typing import TYPE_CHECKING, Any
 
 from agents.model_settings import ModelSettings
 from agents.models.interface import ModelTracing
-from openai.types.responses import (
-    ResponseOutputMessage,
-    ResponseOutputText,
-    ResponseReasoningItem,
-)
+from openai.types.responses import ResponseOutputMessage, ResponseOutputText
 
 from strix.config import load_settings
 from strix.config.models import StrixProvider
@@ -292,22 +288,15 @@ def _checkpoint_item(summary: str) -> dict[str, Any]:
 
 def _extract_text(response: ModelResponse) -> str:
     parts: list[str] = []
-    reasoning: list[str] = []
     for item in response.output:
-        if isinstance(item, ResponseOutputMessage):
-            parts.extend(
-                chunk.text
-                for chunk in item.content
-                if isinstance(chunk, ResponseOutputText) and chunk.text
-            )
-        elif isinstance(item, ResponseReasoningItem):
-            reasoning.extend(summary.text for summary in item.summary if summary.text)
-    if parts or not reasoning:
-        return "".join(parts)
-    # Reasoning models can spend the whole output budget thinking and return no
-    # message; their reasoning is a degraded but usable summary.
-    logger.warning("compaction summary had no output text; using reasoning text")
-    return "\n".join(reasoning)
+        if not isinstance(item, ResponseOutputMessage):
+            continue
+        parts.extend(
+            chunk.text
+            for chunk in item.content
+            if isinstance(chunk, ResponseOutputText) and chunk.text
+        )
+    return "".join(parts)
 
 
 async def _summarize(model: str, prompt: str, max_tokens: int) -> str | None:

strix/skills/__init__.py

@@ -6,6 +6,7 @@ from pathlib import Path
 from typing import TypeGuard
 
 import yaml
+from jinja2 import Environment, StrictUndefined
 
 from strix.utils.resource_paths import get_strix_resource_path
 
@@ -13,6 +14,14 @@ from strix.utils.resource_paths import get_strix_resource_path
 logger = logging.getLogger(__name__)
 
 _FRONTMATTER_PATTERN = re.compile(r"^---\s*\n(?P<body>.*?)\n---\s*\n", re.DOTALL)
+# Only skills with `template: jinja` in their frontmatter are rendered; the rest hold
+# literal `{{ }}` payloads.
+_SKILL_TEMPLATES = Environment(
+    autoescape=False,  # noqa: S701 - prompts, not HTML
+    undefined=StrictUndefined,
+    trim_blocks=True,
+    lstrip_blocks=True,
+)
 
 _INTERNAL_SKILL_CATEGORIES: frozenset[str] = frozenset({"scan_modes", "coordination", "analysis"})
 _ROOT_SKILL_CATEGORY = "root"
@@ -261,7 +270,7 @@ def _candidate_skill_files(skill_name: str) -> list[Path]:
     return _bare_skill_files(skill_name)
 
 
-def load_skills(skill_names: list[str]) -> dict[str, str]:
+def load_skills(skill_names: list[str], *, supports_images: bool = True) -> dict[str, str]:
     """Load skill markdown bodies (frontmatter stripped) by name.
 
     Skill files live at ``strix/skills/<category>/<name>.md`` (or any
@@ -291,7 +300,11 @@ def load_skills(skill_names: list[str]) -> dict[str, str]:
             continue
 
         var_name = skill_name.split("/")[-1]
-        _, skill_body = _parse_skill_content(content, file_path)
+        metadata, skill_body = _parse_skill_content(content, file_path)
+        if metadata.get("template") == "jinja":
+            skill_body = _SKILL_TEMPLATES.from_string(skill_body).render(
+                supports_images=supports_images
+            )
         skill_content[var_name] = skill_body
         logger.debug("Loaded skill: %s -> %s", skill_name, var_name)
         _track_skill_loaded(var_name, file_path)

strix/skills/tooling/agent_browser.md

@@ -1,6 +1,7 @@
 ---
 name: agent_browser
 description: agent-browser CLI for headless Chrome via shell. Snapshot-and-ref workflow, click/fill/extract, screenshots, multi-tab, multi-session, network mocking. Pre-installed in the sandbox; invoke via exec_command.
+template: jinja
 ---
 
 
@@ -260,8 +261,9 @@ only for simple expressions.
 
 ### Screenshot
 
-`agent-browser screenshot` writes a PNG to disk in the sandbox. The
-shell command alone does **not** put the image into your context —
+`agent-browser screenshot` writes a PNG to disk in the sandbox.
+{% if supports_images %}
+The shell command alone does **not** put the image into your context —
 chain it with the SDK ``view_image`` tool to actually see it:
 
 ```bash
@@ -275,6 +277,12 @@ the full path on stdout — pass that to ``view_image``). If you need a
 specific filename, keep it inside that directory or a sibling hidden
 dir under ``/workspace``. Never write screenshots to ``/tmp`` —
 ``view_image`` rejects anything outside the workspace root.
+{% else %}
+You are running on a text-only model and cannot view images. Screenshots
+are only useful to you as a saved file (for example, login evidence).
+Drive the page from `snapshot -i` refs, `eval` for any DOM/JS state you
+need to read, and `text @ref` / `get text` for content extraction.
+{% endif %}
 
 ```bash
 agent-browser screenshot                        # path printed on stdout
@@ -283,6 +291,7 @@ agent-browser screenshot --full                 # full scroll height
 agent-browser screenshot --annotate             # numbered labels + legend keyed to snapshot refs
 ```
 
+{% if supports_images %}
 `--annotate` is designed for multimodal models: each label `[N]` maps
 to ref `@eN`. Take the annotated screenshot, then ``view_image`` it,
 and you can correlate visual layout with snapshot refs.
@@ -294,10 +303,10 @@ layout questions, captchas, custom widgets where the accessibility
 tree is incomplete).
 
 If ``view_image`` errors back at you (rejected image, "vision not
-supported", or similar), you are running on a text-only model — stop
-calling it and stop taking screenshots. Drive the page entirely from
-`snapshot -i` refs, `eval` for any DOM/JS state you need to read, and
-`text @ref` / `get text` for content extraction.
+supported", or similar), you can drive the page from `snapshot -i`
+refs, `eval` for any DOM/JS state you need to read, and `text @ref`
+or `get text` for content extraction.
+{% endif %}
 
 ### Handle multiple pages via tabs
 

strix/tools/load_skill/tool.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 
 from agents import RunContextWrapper, function_tool
 
+from strix.config.models import images_allowed
 from strix.skills import load_skills, validate_requested_skills
 
 
@@ -24,12 +25,11 @@ async def load_skill(ctx: RunContextWrapper, skills: list[str]) -> str:
             Max 5. Names match the bare files under
             ``strix/skills/<category>/<name>.md``.
     """
-    del ctx
     requested = list(skills or [])
     err = validate_requested_skills(requested)
     if err:
         return f"load_skill: {err}"
-    contents = load_skills(requested)
+    contents = load_skills(requested, supports_images=images_allowed(ctx.context))
     if not contents:
         return "load_skill: no content loaded for requested skills."
     sections = [f"## Skill: {name}\n\n{body}" for name, body in contents.items()]

strix/tools/mcp/agent_tools.py

@@ -29,6 +29,7 @@ from typing import TYPE_CHECKING, Any
 
 from agents import RunContextWrapper, function_tool
 
+from strix.config.models import images_allowed
 from strix.tools.mcp.client import _errored_tool_output
 from strix.tools.mcp.naming import namespaced_tool_name
 from strix.tools.mcp.registry import MCP_REGISTRY_CONTEXT_KEY, McpRegistry
@@ -293,4 +294,5 @@ async def call_mcp(
         arguments or {},
         label=namespaced_tool_name(connection, tool),
         result_transform=entry.result_transform,
+        supports_images=images_allowed(ctx.context),
     )

strix/tools/mcp/client.py

@@ -32,6 +32,7 @@ from agents.mcp import (
 )
 from mcp.client.stdio import stdio_client
 from mcp.shared._httpx_utils import create_mcp_http_client
+from mcp.types import TextContent
 
 from strix.tools.mcp.failures import HttpStatusRecorder
 from strix.tools.mcp.session import McpConnectionUnavailableError, SupervisedMcpSession
@@ -183,6 +184,18 @@ def _build_server(config: McpConnectionConfig) -> BuiltMcpServer:
     )
 
 
+def _without_images(result: Any) -> Any:
+    content = [
+        TextContent(
+            type="text", text=f"[{item.mimeType} image omitted: this model cannot view images]"
+        )
+        if item.type == "image"
+        else item
+        for item in result.content
+    ]
+    return result.model_copy(update={"content": content})
+
+
 def _mcp_result_to_tool_output(server: MCPServer, result: Any) -> Any:
     """Serialize a ``CallToolResult`` to a tool output, mirroring the agents SDK.
 
@@ -217,6 +230,7 @@ async def dispatch_mcp_call(
     *,
     label: str,
     result_transform: ResultTransform | None = None,
+    supports_images: bool = True,
 ) -> Any:
     """Run one MCP tool call and convert its result to a tool output.
 
@@ -233,6 +247,8 @@ async def dispatch_mcp_call(
       corrupt the content the agent receives).
     """
     result = await server.call_tool(tool_name, arguments)
+    if not supports_images:
+        result = _without_images(result)
     if result_transform is not None:
         return result_transform(label, result.model_dump(mode="json"))
     tool_output = _mcp_result_to_tool_output(server, result)

strix/tools/mcp/session.py

@@ -365,6 +365,7 @@ class SupervisedMcpSession:
         *,
         label: str,
         result_transform: ResultTransform | None = None,
+        supports_images: bool = True,
     ) -> Any:
         """Run one tool call with bounded retries for transient session failures.
 
@@ -382,6 +383,7 @@ class SupervisedMcpSession:
                 arguments,
                 label=label,
                 result_transform=result_transform,
+                supports_images=supports_images,
             )
 
         outcome = await self._run_job(job, phase="call")

tests/test_agent_factory_shell.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import json
+import threading
 from types import SimpleNamespace
 from typing import Any, cast
 
@@ -10,7 +11,7 @@ import pytest
 from agents.tool import CustomTool, FunctionTool
 
 from strix.agents import factory
-from strix.config import load_settings
+from strix.config import load_settings, models
 
 
 def _capturing_exec_tool(captured: dict[str, str]) -> FunctionTool:
@@ -115,3 +116,18 @@ def test_function_tools_are_result_bounded() -> None:
     by_name = {t.name: t for t in agent.tools}
 
     assert getattr(by_name["think"], "_strix_bounded", False) is True
+
+
+def test_view_image_is_left_out_once_images_are_not_allowed(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    monkeypatch.setattr(models, "_IMAGES_REJECTED", threading.Event())
+    filesystem = factory.build_strix_agent(is_root=True).capabilities[0]
+    filesystem.bind(cast("Any", object()))
+    view_image = next(tool for tool in filesystem.tools() if tool.name == "view_image")
+    is_enabled = cast("Any", cast("FunctionTool", view_image).is_enabled)
+
+    assert is_enabled(SimpleNamespace(context={"supports_images": True}), None) is True
+    assert is_enabled(SimpleNamespace(context={"supports_images": False}), None) is False
+    models.mark_images_rejected()
+    assert is_enabled(SimpleNamespace(context={"supports_images": True}), None) is False

tests/test_compaction.py

@@ -6,8 +6,6 @@ from types import SimpleNamespace
 from typing import TYPE_CHECKING, Any
 
 import pytest
-from agents.extensions.models.litellm_model import InternalChatCompletionMessage
-from agents.models.chatcmpl_converter import Converter
 from litellm.exceptions import BadRequestError, ContextWindowExceededError, RateLimitError
 from openai.types.responses import ResponseOutputMessage, ResponseOutputText
 
@@ -244,29 +242,6 @@ async def test_summarize_routes_through_provider_with_settings(
     assert settings.max_tokens == 64
 
 
-@pytest.mark.asyncio
-async def test_summarize_falls_back_to_reasoning_when_no_output_text(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    _patch_budget(monkeypatch, keep_tokens=30, window=4_000)
-    message = InternalChatCompletionMessage(
-        role="assistant", content=None, reasoning_content="REASONED SUMMARY"
-    )
-    response = SimpleNamespace(output=Converter.message_to_output_items(message))
-
-    class FakeModel:
-        async def get_response(self, **_kwargs: Any) -> Any:
-            return response
-
-    class FakeProvider:
-        def get_model(self, _model_name: str | None) -> Any:
-            return FakeModel()
-
-    monkeypatch.setattr(compaction, "StrixProvider", FakeProvider)
-
-    assert await compaction._summarize("m", "p", 64) == "REASONED SUMMARY"
-
-
 def test_fit_to_tokens_truncates_oversized_text(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(compaction, "count_tokens", lambda _m, t: len(t))
     text = "x" * 10_000

tests/test_execution.py

@@ -9,9 +9,11 @@ from typing import Any, cast
 from unittest.mock import MagicMock
 
 import pytest
+from agents import RunConfig
 from agents.exceptions import MaxTurnsExceeded
 from agents.items import MessageOutputItem
 from agents.memory import SQLiteSession
+from agents.run_config import CallModelData, ModelInputData
 from agents.tool_context import ToolContext
 from openai.types.responses import ResponseOutputMessage, ResponseOutputRefusal, ResponseOutputText
 
@@ -1531,3 +1533,32 @@ async def test_autonomous_nudge_does_not_offer_the_user() -> None:
     )
 
     assert "wait_for_user" not in items[0]["content"]
+
+
+@pytest.mark.asyncio
+async def test_text_only_filter_scrubs_images_then_chains_existing_filter() -> None:
+    run_config = RunConfig(model="m")
+    assert execution._with_image_scrub(run_config, {"supports_images": True}) is run_config
+
+    seen: list[ModelInputData] = []
+
+    def _inner(data: CallModelData[Any]) -> ModelInputData:
+        seen.append(data.model_data)
+        return data.model_data
+
+    run_config = RunConfig(model="m", call_model_input_filter=_inner)
+    scrub: Any = execution._with_image_scrub(run_config, {"supports_images": False})
+    image = {"type": "input_image", "image_url": "data:image/png;base64,aGk="}
+    output = {"type": "function_call_output", "call_id": "c1", "output": [image]}
+    data = CallModelData(
+        model_data=ModelInputData(input=[cast("Any", output)], instructions="sys"),
+        agent=MagicMock(),
+        context=None,
+    )
+
+    result = await scrub.call_model_input_filter(data)
+    assert result is seen[0]
+    assert result.input[0]["output"] == [
+        {"type": "input_text", "text": execution._TEXT_ONLY_IMAGE_TEXT}
+    ]
+    assert data.model_data.input[0]["output"] == [image]

tests/test_execution_transient_retry.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
-from typing import Any, cast
+import threading
+from typing import Any
 
 import httpx
 import pytest
@@ -15,7 +16,7 @@ from openai import (
     RateLimitError,
 )
 
-from strix.config import codex
+from strix.config import codex, models
 from strix.core import execution
 from strix.core.agents import AgentCoordinator
 
@@ -109,6 +110,7 @@ def _patch_fast_backoff(monkeypatch: pytest.MonkeyPatch) -> None:
 async def _run_once(
     monkeypatch: pytest.MonkeyPatch,
     streams: list[_FakeStream],
+    session: Any = None,
 ) -> Any:
     _patch_fast_backoff(monkeypatch)
     calls = {"n": 0}
@@ -128,10 +130,10 @@ async def _run_once(
         coordinator,
         "root",
         input_data="task",
-        run_config=cast("RunConfig", object()),
+        run_config=RunConfig(),
         context={},
         max_turns=5,
-        session=None,
+        session=session,
         interactive=False,
         event_sink=None,
         hooks=None,
@@ -172,3 +174,76 @@ async def test_run_cycle_does_not_retry_permanent_error(
     streams = [_FakeStream(exc=bad_request), _FakeStream()]
     with pytest.raises(BadRequestError):
         await _run_once(monkeypatch, streams)
+
+
+@pytest.mark.parametrize(
+    ("status", "message", "expected"),
+    [
+        (
+            404,
+            "litellm.NotFoundError: NotFoundError: OpenrouterException - "
+            '{"error":{"message":"No endpoints found that support image input","code":404,'
+            '"metadata":{"failed_routing_step":"Filter by Image Support"}}}',
+            True,
+        ),
+        (400, "Invalid content type. image_url is only supported by certain models.", True),
+        (400, "my-model is not a multimodal model", True),
+        (400, "This model doesn't support the image field for user messages.", True),
+        (
+            400,
+            "OpenRouterException: Message: This model's maximum context length is 1048576 "
+            "tokens, but the request requires 1073484 tokens (942412 input including "
+            "image/vision expansion + 131072 for the completion). Reduce the input length "
+            "or max_tokens.",
+            False,
+        ),
+        (400, "At most 5 image(s) may be provided in one prompt.", False),
+        (
+            400,
+            "Upstream error from Relace: Generation stopped with an incomplete tool call",
+            False,
+        ),
+        (500, "No endpoints found that support image input", False),
+    ],
+)
+def test_is_image_rejection(status: int, message: str, expected: bool) -> None:
+    exc = APIStatusError(message, response=httpx.Response(status, request=_request()), body=None)
+    assert execution._is_image_rejection(exc) is expected
+
+
+async def _run_with_image_session(
+    monkeypatch: pytest.MonkeyPatch, streams: list[_FakeStream]
+) -> tuple[Any, int, int]:
+    strips = {"n": 0}
+
+    async def _fake_strip(_session: Any) -> bool:
+        strips["n"] += 1
+        return True
+
+    async def _no_compact(*_args: Any, **_kwargs: Any) -> bool:
+        return False
+
+    monkeypatch.setattr(execution, "strip_all_images_from_session", _fake_strip)
+    monkeypatch.setattr(execution, "_compact_session", _no_compact)
+    result, attempts, _coordinator = await _run_once(monkeypatch, streams, session=object())
+    return result, attempts, strips["n"]
+
+
+@pytest.mark.asyncio
+async def test_run_cycle_strips_images_on_image_rejection(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    monkeypatch.setattr(models, "_IMAGES_REJECTED", threading.Event())
+    rejection = APIStatusError(
+        "No endpoints found that support image input",
+        response=httpx.Response(status_code=404, request=_request()),
+        body=None,
+    )
+    streams = [_FakeStream(exc=rejection), _FakeStream()]
+    result, attempts, strips = await _run_with_image_session(monkeypatch, streams)
+
+    assert result is streams[1]
+    assert attempts == 2
+    assert strips == 1
+    # Every later request in the run, from any agent, has its images scrubbed.
+    assert execution._with_image_scrub(RunConfig(), {}).call_model_input_filter is not None

tests/test_mcp_client.py

@@ -18,7 +18,7 @@ from typing import TYPE_CHECKING, Any
 import pytest
 from agents.mcp import MCPServer, MCPServerStdio, MCPServerStreamableHttp
 from agents.tool_context import ToolContext
-from mcp.types import CallToolResult, TextContent
+from mcp.types import CallToolResult, ImageContent, TextContent
 from mcp.types import Tool as MCPTool
 from pydantic import ValidationError
 
@@ -884,6 +884,51 @@ async def test_call_mcp_flags_an_errored_result_failed_for_the_tui() -> None:
     assert out == {"type": "text", "text": "boom:read_file", "success": False}
 
 
+@pytest.mark.asyncio
+async def test_call_mcp_replaces_images_for_a_text_only_model() -> None:
+    class _ImageServer(FakeMCPServer):
+        async def call_tool(
+            self,
+            tool_name: str,
+            arguments: dict[str, Any] | None,
+            meta: dict[str, Any] | None = None,
+        ) -> CallToolResult:
+            return CallToolResult(
+                content=[
+                    TextContent(type="text", text="shot"),
+                    ImageContent(type="image", data="aGk=", mimeType="image/png"),
+                ]
+            )
+
+    registry = McpRegistry()
+    registry.add(
+        name="fs", server=_ImageServer("fs", [_mcp_tool("shot")]), purpose=None, tool_count=1
+    )
+    ctx = _ctx(registry)
+    args = json.dumps({"connection": "fs", "tool": "shot"})
+
+    out = await call_mcp.on_invoke_tool(ctx, args)
+    assert out[1] == {"type": "image", "image_url": "data:image/png;base64,aGk="}
+
+    ctx.context["supports_images"] = False
+    out = await call_mcp.on_invoke_tool(ctx, args)
+    assert out == [
+        {"type": "text", "text": "shot"},
+        {"type": "text", "text": "[image/png image omitted: this model cannot view images]"},
+    ]
+
+    transformed: list[Any] = []
+    registry.add(
+        name="pro",
+        server=_ImageServer("pro", [_mcp_tool("shot")]),
+        purpose=None,
+        tool_count=1,
+        result_transform=lambda _label, result: transformed.append(result),
+    )
+    await call_mcp.on_invoke_tool(ctx, json.dumps({"connection": "pro", "tool": "shot"}))
+    assert [block["type"] for block in transformed[0]["content"]] == ["text", "text"]
+
+
 # --- generic MCP tools are the only MCP surface every agent gets -------------
 
 

tests/test_models.py

@@ -16,6 +16,7 @@ from strix.config.models import (
     _NonStreamingModel,
     _TurnGuardModel,
     configure_sdk_model_defaults,
+    model_supports_images,
     request_timeout_extra_args,
     resolve_api_type,
     routes_through_litellm,
@@ -202,3 +203,11 @@ def test_configure_sdk_api_route_follows_the_given_model(
     models.configure_sdk_api_route("my-private-model", settings)
 
     assert routes == ["responses", "chat_completions"]
+
+
+def test_image_support_follows_the_catalog(monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setitem(litellm.model_cost, "acme-text", {})
+    monkeypatch.setitem(litellm.model_cost, "acme-vision", {"supports_vision": True})
+    assert model_supports_images("acme-text") is False
+    assert model_supports_images("litellm/openai/acme-vision") is True
+    assert model_supports_images("acme-unknown-model") is True

tests/test_runner_root_prompt.py

@@ -13,6 +13,7 @@ from typing import Any
 import httpx
 import pytest
 from agents import ModelSettings
+from agents.tool_context import ToolContext
 from openai import RateLimitError
 
 import strix.tools.mcp as mcp_pkg
@@ -24,6 +25,7 @@ from strix.core import runner
 from strix.core.agents import AgentCoordinator
 from strix.core.inputs import make_model_settings
 from strix.runtime import session_manager
+from strix.tools.load_skill.tool import load_skill
 from strix.tools.mcp import BearerAuth, McpConnectionConfig, McpConnectionRequest
 
 
@@ -287,6 +289,29 @@ def test_requested_skills_follow_the_shared_prefix() -> None:
     assert "<xss>" in xss.split("<cache_point>")[1]
 
 
+def test_text_only_prompt_drops_screenshot_guidance() -> None:
+    assert "view_image" in render_system_prompt(include_scope=False)
+
+    prompt = render_system_prompt(include_scope=False, supports_images=False)
+    assert "view_image" not in prompt
+    assert "text-only model and cannot view images" in prompt
+    assert "<!--" not in prompt
+
+
+@pytest.mark.asyncio
+async def test_text_only_load_skill_drops_screenshot_guidance() -> None:
+    ctx = ToolContext(
+        context={"supports_images": False},
+        tool_name="load_skill",
+        tool_call_id="call-1",
+        tool_arguments="{}",
+    )
+
+    out = await load_skill.on_invoke_tool(ctx, '{"skills": ["agent_browser"]}')
+    assert "view_image" not in out
+    assert "text-only model" in out
+
+
 def test_scope_is_sent_as_its_own_system_message_on_cache_point_routes() -> None:
     settings = make_model_settings(None, model_name="anthropic/claude-sonnet-5-5")
     prompt = render_system_prompt(

tests/test_skill_dir_extension.py

@@ -70,6 +70,17 @@ def test_registered_dir_adds_new_skill(tmp_path: Path) -> None:
     assert load_skills(["widget"]) == {"widget": "widget body"}
 
 
+def test_only_jinja_skills_are_rendered(tmp_path: Path) -> None:
+    body = "{% if supports_images %}\nlook\n{% else %}\nread\n{% endif %}\n"
+    _write_skill(tmp_path, "extra", "plain", body)
+    _write_skill(tmp_path, "extra", "templated", f"---\ntemplate: jinja\n---\n{body}")
+    register_skill_dir(tmp_path)
+
+    assert load_skills(["plain"]) == {"plain": body}
+    assert load_skills(["templated"]) == {"templated": "look\n"}
+    assert load_skills(["templated"], supports_images=False) == {"templated": "read\n"}
+
+
 def test_available_skill_includes_frontmatter_description(tmp_path: Path) -> None:
     _write_skill(
         tmp_path,