docagent-cli 0.0.40__tar.gz → 0.0.42__tar.gz

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## [0.0.42](https://github.com/langchain-ai/docagent/compare/docagent-cli==0.0.41...docagent-cli==0.0.42) (2026-04-10)
+
+### Bug Fixes
+
+* Fix `JSONEncoder` crash caused by `encoding="utf-8"` being incorrectly injected into `json.dumps()` calls instead of `write_text()` during the 0.0.41 encoding fix
+
+## [0.0.41](https://github.com/langchain-ai/docagent/compare/docagent-cli==0.0.40...docagent-cli==0.0.41) (2026-04-10)
+
+### Bug Fixes
+
+* Fix `UnicodeDecodeError` on Chinese Windows systems (GBK locale) by explicitly specifying `encoding="utf-8"` on all `read_text()` and `write_text()` file operations ([#fix-gbk-encoding](https://github.com/langchain-ai/docagent/issues/fix-gbk-encoding))
+
 ## [0.0.34](https://github.com/langchain-ai/docagent/compare/docagent-cli==0.0.33...docagent-cli==0.0.34) (2026-03-17)
 
 ### Features

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docagent-cli
-Version: 0.0.40
+Version: 0.0.42
 Summary: Terminal interface for Deep Agents - interactive AI agent with file operations, shell access, and sub-agent capabilities.
 Project-URL: Homepage, https://docs.langchain.com/oss/python/docagent/overview
 Project-URL: Documentation, https://reference.langchain.com/python/docagent/

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/docagent_cli/_version.py

@@ -1,6 +1,6 @@
 """Version information and lightweight constants for `docagent-cli`."""
 
-__version__ = "0.0.40"  # x-release-please-version
+__version__ = "0.0.42"  # x-release-please-version
 
 DOCS_URL = "https://docs.langchain.com/oss/python/docagent/cli"
 """URL for `docagent-cli` documentation."""

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/docagent_cli/agent.py

@@ -365,7 +365,7 @@ def reset_agent(
             )
             raise SystemExit(1)
 
-        source_content = source_md.read_text()
+        source_content = source_md.read_text(encoding="utf-8")
         action_desc = f"contents of agent '{source_agent}'"
     else:
         source_content = get_default_coding_instructions()
@@ -399,7 +399,7 @@ def reset_agent(
 
     agent_dir.mkdir(parents=True, exist_ok=True)
     agent_md = agent_dir / "AGENTS.md"
-    agent_md.write_text(source_content)
+    agent_md.write_text(source_content, encoding="utf-8")
 
     if output_format == "json":
         from docagent_cli.output import write_json
@@ -504,7 +504,7 @@ def get_system_prompt(
         ... {CONDITIONAL SECTIONS} ...
         ```
     """
-    template = (Path(__file__).parent / "system_prompt.md").read_text()
+    template = (Path(__file__).parent / "system_prompt.md").read_text(encoding="utf-8")
 
     skills_path = f"~/.docagent/{assistant_id}/skills"
 

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/docagent_cli/built_in_skills/docx/scripts/accept_changes.py

@@ -92,7 +92,7 @@ def _setup_libreoffice_macro() -> bool:
     macro_dir = Path(MACRO_DIR)
     macro_file = macro_dir / "Module1.xba"
 
-    if macro_file.exists() and "AcceptAllTrackedChanges" in macro_file.read_text():
+    if macro_file.exists() and "AcceptAllTrackedChanges" in macro_file.read_text(encoding="utf-8"):
         return True
 
     if not macro_dir.exists():
@@ -111,7 +111,7 @@ def _setup_libreoffice_macro() -> bool:
         macro_dir.mkdir(parents=True, exist_ok=True)
 
     try:
-        macro_file.write_text(ACCEPT_CHANGES_MACRO)
+        macro_file.write_text(ACCEPT_CHANGES_MACRO, encoding="utf-8")
         return True
     except Exception as e:
         logger.warning(f"Failed to setup LibreOffice macro: {e}")

{docagent_cli-0.0.40/docagent_cli/built_in_skills/pptx → docagent_cli-0.0.42/docagent_cli/built_in_skills/docx}/scripts/office/soffice.py

@@ -55,7 +55,7 @@ def _ensure_shim() -> Path:
         return _SHIM_SO
 
     src = Path(tempfile.gettempdir()) / "lo_socket_shim.c"
-    src.write_text(_SHIM_SOURCE)
+    src.write_text(_SHIM_SOURCE, encoding="utf-8")
     subprocess.run(
         ["gcc", "-shared", "-fPIC", "-o", str(_SHIM_SO), str(src), "-ldl"],
         check=True,

docagent_cli-0.0.42/docagent_cli/built_in_skills/long-doc-writing/SKILL.md

@@ -0,0 +1,137 @@
+---
+name: long-doc-writing
+description: Guide for writing long-form documents — research papers, technical reports, literature reviews, methodology sections — with natural prose style. Use this skill when the user asks to write, draft, expand, or polish any substantial document section, especially academic or research writing. Trigger whenever the user wants to write about evaluation metrics, methodology, results, related work, or any multi-paragraph research content. Strongly prefer this skill over plain generation whenever the user mentions writing that should feel natural, narrative, or scholarly rather than bullet-pointed or template-driven.
+---
+
+# Long-Form Document Writing
+
+This skill guides the production of long-form documents — research papers, technical reports, literature reviews, methodology sections, and similar scholarly or professional writing — with a focus on prose quality over mechanical structure.
+
+The core philosophy: good long-form writing reads like a senior researcher thinking through a problem out loud, not like a checklist being filled in. Every section should have a clear reason for existing, every sentence should carry information, and the transitions between ideas should feel earned.
+
+## Foundational Principles
+
+### Prose Over Lists
+
+When content appears as a list — bullet points, numbered items, isolated definitions — convert it into flowing paragraphs. This is not cosmetic. Lists fragment reasoning; paragraphs let ideas build on each other and reveal relationships that bullets obscure.
+
+**Instead of:**
+> The evaluation metrics include:
+> - BLEU score: measures n-gram overlap
+> - ROUGE-L: measures longest common subsequence
+> - BERTScore: measures semantic similarity
+
+**Write:**
+> To assess generation quality across its different dimensions, we rely on three metrics that together capture what simple accuracy cannot. BLEU score grounds the evaluation in surface-level fidelity — if the model's output shares n-gram patterns with the reference, it is at minimum producing plausible text. But n-gram overlap alone misses the forest for the trees, which is why we pair it with ROUGE-L: the longest common subsequence it tracks tends to surface whether the model captured the right information in roughly the right order, even if the phrasing differs. Neither metric, however, can tell us whether two semantically equivalent sentences are actually equivalent — for that, BERTScore fills the gap, using contextual embeddings to measure meaning-level similarity rather than surface form.
+
+This example illustrates the key moves: each metric gets a *reason*, not just a definition, and the paragraph builds from one idea to the next rather than presenting three isolated facts.
+
+### Narrative Integration of Technical Content
+
+Every technical element — a metric, a design choice, a baseline, a dataset — should be introduced with a brief explanation of *why it belongs in this document*. Not "Metric X is defined as Y", but "We chose metric X because it captures Z, which matters here because..."
+
+This is especially important for:
+- **Evaluation metrics**: Explain what each metric is designed to detect and why that matters for your specific task.
+- **Baselines and comparisons**: Explain why each comparison is informative, not just what it is.
+- **Datasets**: Explain what properties make the dataset appropriate, not just its size and source.
+- **Model components**: Explain what problem each component solves, not just what it does.
+
+### Active Voice and Researcher's Tone
+
+Write as if a knowledgeable colleague is explaining their thinking — confident, direct, sometimes opinionated. Favor active constructions:
+
+| Passive (avoid) | Active (prefer) |
+|---|---|
+| "BLEU score was used to evaluate..." | "We evaluate using BLEU score because..." |
+| "It can be observed that..." | "We find that..." |
+| "The model was trained on..." | "We trained the model on..." |
+| "Results are presented in Table 2" | "Table 2 shows..." |
+
+This is not about being casual; it is about being clear. Active voice is also shorter and harder to misread.
+
+### No Template Boilerplate
+
+Avoid clichéd academic openers and transitions that carry no information:
+
+**Phrases to eliminate:**
+- "The evaluation metrics include..."
+- "In this section, we present..."
+- "It is worth noting that..."
+- "As can be seen from the results..."
+- "We now turn to..."
+- "For the purpose of this study..."
+- "It is important to mention that..."
+- "This paper is organized as follows..."
+
+When you catch yourself writing one of these, ask: what am I actually trying to say? Then say that instead.
+
+---
+
+## Section-Specific Guidance
+
+### Evaluation / Metrics Sections
+
+This is where mechanical writing is most common and most damaging. The typical failure mode: list each metric with a one-line definition and a formula. The reader learns what the metrics are but not *why this paper uses them* or *what the results actually mean*.
+
+**Structure to aim for:**
+1. Open by naming the core question the evaluation is trying to answer — not "we evaluate the model" but "we want to know whether the model preserves factual accuracy while improving fluency."
+2. Introduce metrics in the order they address that question, explaining the gap each one fills.
+3. Note limitations of your metrics honestly — this shows methodological awareness and actually strengthens the paper.
+4. Connect metrics to results in the same breath where possible: "We track ROUGE-L because it captures order-sensitive recall — and as Table 3 shows, this is precisely where our model most consistently outperforms baselines."
+
+### Methodology / Approach Sections
+
+The goal is to let readers reconstruct your work, but also to understand your reasoning. Every major decision should come with its rationale:
+
+- Why this architecture over alternatives?
+- Why this training split?
+- Why this preprocessing step?
+
+Anticipate the skeptical reader. If you made a choice that might seem arbitrary, explain why it was not.
+
+### Related Work Sections
+
+The most common failure: a list of papers with one-sentence summaries, no synthesis. Good related work sections tell a story about the field:
+
+- What problem was everyone trying to solve?
+- How did approaches evolve?
+- Where does your work fit in that trajectory?
+- What does your work do that prior work cannot?
+
+Group papers by *what they do* or *what gap they address*, not by chronology or who wrote them.
+
+### Results and Discussion Sections
+
+Do not just describe tables — analyze them. "Table 2 shows our method achieves 87.3 BLEU" is a caption, not an analysis. A results section should answer:
+
+- What do the numbers mean, qualitatively?
+- Where does the method succeed and why?
+- Where does it fall short and why?
+- What surprising or counterintuitive result emerged, and what does it tell you?
+
+---
+
+## Editing Pass Checklist
+
+After drafting, run through these questions before finalizing any section:
+
+1. **Is there a list that could be a paragraph?** Convert it.
+2. **Does every metric / baseline / dataset have a reason for being here?** Add it if not.
+3. **Does any sentence start with "It is" or "There are"?** Revise to lead with the subject.
+4. **Are there transition phrases that add no information?** Delete them.
+5. **Does each paragraph have a clear point?** If you cannot summarize a paragraph in one sentence, it probably needs restructuring.
+6. **Is the reader told not just *what* but *why*?** Wherever the answer is no, add the reasoning.
+7. **Could a colleague from a neighboring field understand the stakes?** If not, the motivation is probably buried in jargon.
+
+---
+
+## Working With the User
+
+When the user provides content to write or rewrite:
+
+1. **Identify the core argument or finding** that the section is trying to communicate. If it is unclear, ask.
+2. **Draft in prose**, applying the principles above. Do not produce bullet points unless the user explicitly asks for them.
+3. **Surface one or two decisions** you made that the user might want to revisit — e.g., "I grouped metrics by what they measure rather than the order you listed them; let me know if you'd prefer a different organization."
+4. **Invite targeted feedback** rather than asking "does this look good?" — e.g., "Does the tone feel right for your venue? Should the metrics section lead with the primary metric or with the research question it answers?"
+
+If the user provides a list of points they want covered, treat it as raw material, not a structure. Your job is to find the narrative thread and write toward it.

{docagent_cli-0.0.40/docagent_cli/built_in_skills/docx → docagent_cli-0.0.42/docagent_cli/built_in_skills/pptx}/scripts/office/soffice.py

@@ -55,7 +55,7 @@ def _ensure_shim() -> Path:
         return _SHIM_SO
 
     src = Path(tempfile.gettempdir()) / "lo_socket_shim.c"
-    src.write_text(_SHIM_SOURCE)
+    src.write_text(_SHIM_SOURCE, encoding="utf-8")
     subprocess.run(
         ["gcc", "-shared", "-fPIC", "-o", str(_SHIM_SO), str(src), "-ldl"],
         check=True,

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/docagent_cli/built_in_skills/skill-creator/eval-viewer/generate_review.py

@@ -24,7 +24,7 @@ import sys
 import time
 import webbrowser
 from functools import partial
-from http.server import HTTPServer, BaseHTTPRequestHandler
+from http.server import BaseHTTPRequestHandler, HTTPServer
 from pathlib import Path
 
 # Files to exclude from output listings
@@ -32,9 +32,32 @@ METADATA_FILES = {"transcript.md", "user_notes.md", "metrics.json"}
 
 # Extensions we render as inline text
 TEXT_EXTENSIONS = {
-    ".txt", ".md", ".json", ".csv", ".py", ".js", ".ts", ".tsx", ".jsx",
-    ".yaml", ".yml", ".xml", ".html", ".css", ".sh", ".rb", ".go", ".rs",
-    ".java", ".c", ".cpp", ".h", ".hpp", ".sql", ".r", ".toml",
+    ".txt",
+    ".md",
+    ".json",
+    ".csv",
+    ".py",
+    ".js",
+    ".ts",
+    ".tsx",
+    ".jsx",
+    ".yaml",
+    ".yml",
+    ".xml",
+    ".html",
+    ".css",
+    ".sh",
+    ".rb",
+    ".go",
+    ".rs",
+    ".java",
+    ".c",
+    ".cpp",
+    ".h",
+    ".hpp",
+    ".sql",
+    ".r",
+    ".toml",
 }
 
 # Extensions we render as inline images
@@ -88,10 +111,13 @@ def build_run(root: Path, run_dir: Path) -> dict | None:
     eval_id = None
 
     # Try eval_metadata.json
-    for candidate in [run_dir / "eval_metadata.json", run_dir.parent / "eval_metadata.json"]:
+    for candidate in [
+        run_dir / "eval_metadata.json",
+        run_dir.parent / "eval_metadata.json",
+    ]:
         if candidate.exists():
             try:
-                metadata = json.loads(candidate.read_text())
+                metadata = json.loads(candidate.read_text(encoding="utf-8"))
                 prompt = metadata.get("prompt", "")
                 eval_id = metadata.get("eval_id")
             except (json.JSONDecodeError, OSError):
@@ -101,10 +127,13 @@ def build_run(root: Path, run_dir: Path) -> dict | None:
 
     # Fall back to transcript.md
     if not prompt:
-        for candidate in [run_dir / "transcript.md", run_dir / "outputs" / "transcript.md"]:
+        for candidate in [
+            run_dir / "transcript.md",
+            run_dir / "outputs" / "transcript.md",
+        ]:
             if candidate.exists():
                 try:
-                    text = candidate.read_text()
+                    text = candidate.read_text(encoding="utf-8")
                     match = re.search(r"## Eval Prompt\n\n([\s\S]*?)(?=\n##|$)", text)
                     if match:
                         prompt = match.group(1).strip()
@@ -131,7 +160,7 @@ def build_run(root: Path, run_dir: Path) -> dict | None:
     for candidate in [run_dir / "grading.json", run_dir.parent / "grading.json"]:
         if candidate.exists():
             try:
-                grading = json.loads(candidate.read_text())
+                grading = json.loads(candidate.read_text(encoding="utf-8"))
             except (json.JSONDecodeError, OSError):
                 pass
             if grading:
@@ -166,7 +195,11 @@ def embed_file(path: Path) -> dict:
             raw = path.read_bytes()
             b64 = base64.b64encode(raw).decode("ascii")
         except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
+            return {
+                "name": path.name,
+                "type": "error",
+                "content": "(Error reading file)",
+            }
         return {
             "name": path.name,
             "type": "image",
@@ -178,7 +211,11 @@ def embed_file(path: Path) -> dict:
             raw = path.read_bytes()
             b64 = base64.b64encode(raw).decode("ascii")
         except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
+            return {
+                "name": path.name,
+                "type": "error",
+                "content": "(Error reading file)",
+            }
         return {
             "name": path.name,
             "type": "pdf",
@@ -189,7 +226,11 @@ def embed_file(path: Path) -> dict:
             raw = path.read_bytes()
             b64 = base64.b64encode(raw).decode("ascii")
         except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
+            return {
+                "name": path.name,
+                "type": "error",
+                "content": "(Error reading file)",
+            }
         return {
             "name": path.name,
             "type": "xlsx",
@@ -201,7 +242,11 @@ def embed_file(path: Path) -> dict:
             raw = path.read_bytes()
             b64 = base64.b64encode(raw).decode("ascii")
         except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
+            return {
+                "name": path.name,
+                "type": "error",
+                "content": "(Error reading file)",
+            }
         return {
             "name": path.name,
             "type": "binary",
@@ -222,7 +267,7 @@ def load_previous_iteration(workspace: Path) -> dict[str, dict]:
     feedback_path = workspace / "feedback.json"
     if feedback_path.exists():
         try:
-            data = json.loads(feedback_path.read_text())
+            data = json.loads(feedback_path.read_text(encoding="utf-8"))
             feedback_map = {
                 r["run_id"]: r["feedback"]
                 for r in data.get("reviews", [])
@@ -255,7 +300,7 @@ def generate_html(
 ) -> str:
     """Generate the complete standalone HTML page with embedded data."""
     template_path = Path(__file__).parent / "viewer.html"
-    template = template_path.read_text()
+    template = template_path.read_text(encoding="utf-8")
 
     # Build previous_feedback and previous_outputs maps for the template
     previous_feedback: dict[str, str] = {}
@@ -278,19 +323,24 @@ def generate_html(
 
     data_json = json.dumps(embedded)
 
-    return template.replace("/*__EMBEDDED_DATA__*/", f"const EMBEDDED_DATA = {data_json};")
+    return template.replace(
+        "/*__EMBEDDED_DATA__*/", f"const EMBEDDED_DATA = {data_json};"
+    )
 
 
 # ---------------------------------------------------------------------------
 # HTTP server (stdlib only, zero dependencies)
 # ---------------------------------------------------------------------------
 
+
 def _kill_port(port: int) -> None:
     """Kill any process listening on the given port."""
     try:
         result = subprocess.run(
             ["lsof", "-ti", f":{port}"],
-            capture_output=True, text=True, timeout=5,
+            capture_output=True,
+            text=True,
+            timeout=5,
         )
         for pid_str in result.stdout.strip().split("\n"):
             if pid_str.strip():
@@ -305,6 +355,7 @@ def _kill_port(port: int) -> None:
     except FileNotFoundError:
         print("Note: lsof not found, cannot check if port is in use", file=sys.stderr)
 
+
 class ReviewHandler(BaseHTTPRequestHandler):
     """Serves the review HTML and handles feedback saves.
 
@@ -336,7 +387,9 @@ class ReviewHandler(BaseHTTPRequestHandler):
             benchmark = None
             if self.benchmark_path and self.benchmark_path.exists():
                 try:
-                    benchmark = json.loads(self.benchmark_path.read_text())
+                    benchmark = json.loads(
+                        self.benchmark_path.read_text(encoding="utf-8")
+                    )
                 except (json.JSONDecodeError, OSError):
                     pass
             html = generate_html(runs, self.skill_name, self.previous, benchmark)
@@ -366,7 +419,9 @@ class ReviewHandler(BaseHTTPRequestHandler):
                 data = json.loads(body)
                 if not isinstance(data, dict) or "reviews" not in data:
                     raise ValueError("Expected JSON object with 'reviews' key")
-                self.feedback_path.write_text(json.dumps(data, indent=2) + "\n")
+                self.feedback_path.write_text(
+                    json.dumps(data, indent=2) + "\n", encoding="utf-8"
+                )
                 resp = b'{"ok":true}'
                 self.send_response(200)
             except (json.JSONDecodeError, OSError, ValueError) as e:
@@ -387,18 +442,29 @@ class ReviewHandler(BaseHTTPRequestHandler):
 def main() -> None:
     parser = argparse.ArgumentParser(description="Generate and serve eval review")
     parser.add_argument("workspace", type=Path, help="Path to workspace directory")
-    parser.add_argument("--port", "-p", type=int, default=3117, help="Server port (default: 3117)")
-    parser.add_argument("--skill-name", "-n", type=str, default=None, help="Skill name for header")
     parser.add_argument(
-        "--previous-workspace", type=Path, default=None,
+        "--port", "-p", type=int, default=3117, help="Server port (default: 3117)"
+    )
+    parser.add_argument(
+        "--skill-name", "-n", type=str, default=None, help="Skill name for header"
+    )
+    parser.add_argument(
+        "--previous-workspace",
+        type=Path,
+        default=None,
         help="Path to previous iteration's workspace (shows old outputs and feedback as context)",
     )
     parser.add_argument(
-        "--benchmark", type=Path, default=None,
+        "--benchmark",
+        type=Path,
+        default=None,
         help="Path to benchmark.json to show in the Benchmark tab",
     )
     parser.add_argument(
-        "--static", "-s", type=Path, default=None,
+        "--static",
+        "-s",
+        type=Path,
+        default=None,
         help="Write standalone HTML to this path instead of starting a server",
     )
     args = parser.parse_args()
@@ -424,21 +490,23 @@ def main() -> None:
     benchmark = None
     if benchmark_path and benchmark_path.exists():
         try:
-            benchmark = json.loads(benchmark_path.read_text())
+            benchmark = json.loads(benchmark_path.read_text(encoding="utf-8"))
         except (json.JSONDecodeError, OSError):
             pass
 
     if args.static:
         html = generate_html(runs, skill_name, previous, benchmark)
         args.static.parent.mkdir(parents=True, exist_ok=True)
-        args.static.write_text(html)
+        args.static.write_text(html, encoding="utf-8")
         print(f"\n  Static viewer written to: {args.static}\n")
         sys.exit(0)
 
     # Kill any existing process on the target port
     port = args.port
     _kill_port(port)
-    handler = partial(ReviewHandler, workspace, skill_name, feedback_path, previous, benchmark_path)
+    handler = partial(
+        ReviewHandler, workspace, skill_name, feedback_path, previous, benchmark_path
+    )
     try:
         server = HTTPServer(("127.0.0.1", port), handler)
     except OSError:

{docagent_cli-0.0.40 → docagent_cli-0.0.42}/docagent_cli/built_in_skills/skill-creator/scripts/generate_report.py

@@ -311,12 +311,12 @@ def main():
     if args.input == "-":
         data = json.load(sys.stdin)
     else:
-        data = json.loads(Path(args.input).read_text())
+        data = json.loads(Path(args.input).read_text(encoding="utf-8"))
 
     html_output = generate_html(data, skill_name=args.skill_name)
 
     if args.output:
-        Path(args.output).write_text(html_output)
+        Path(args.output).write_text(html_output, encoding="utf-8")
         print(f"Report written to {args.output}", file=sys.stderr)
     else:
         print(html_output)