kolega-code 0.14.0__tar.gz → 0.15.0__tar.gz

{kolega_code-0.14.0 → kolega_code-0.15.0}/CHANGELOG.md

@@ -6,6 +6,22 @@ This project uses GitHub Releases for detailed generated release notes. This fil
 
 ## Unreleased
 
+## 0.15.0 - 2026-07-07
+
+### Added
+
+- Added Language Server Protocol (LSP) integration with an agent-callable
+  read-only `lsp` tool (diagnostics, go-to-definition, references, hover,
+  document/workspace symbols, and status) and a trusted `lsp_edit` tool for
+  rename, formatting, and code actions. The `edit`, `multi_edit`, and `write`
+  tools now append LSP diagnostics to their results, with a `/lsp` status
+  command, detected-language status in the TUI wordmark, and a Settings toggle.
+- Added bounded skill metadata rendering with a context-aware token budget.
+  The prompt catalog now uses skill name and description only, descriptions are
+  truncated before skills are omitted, and `list_skills` is queryable with
+  `max_results`.
+- Documented composable workflow shapes in the Gigacode orchestration guide.
+
 ## 0.14.0 - 2026-07-03
 
 ### Added

{kolega_code-0.14.0 → kolega_code-0.15.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kolega-code
-Version: 0.14.0
+Version: 0.15.0
 Summary: Multi-agent coding in the terminal
 Project-URL: Homepage, https://kolega-ai.github.io/kolega-code/
 Project-URL: Documentation, https://kolega-ai.github.io/kolega-code/

{kolega_code-0.14.0 → kolega_code-0.15.0}/docs/astro.config.mjs

@@ -68,6 +68,10 @@ export default defineConfig({
               label: "MCP Servers",
               slug: "configuration/mcp",
             },
+            {
+              label: "Language Servers",
+              slug: "configuration/lsp",
+            },
             {
               label: "Environment Variables",
               slug: "configuration/environment-variables",

{kolega_code-0.14.0 → kolega_code-0.15.0}/docs/src/assets/coverage.svg

@@ -1 +1 @@
-<svg xmlns="http://www.w3.org/2000/svg" width="114" height="20" role="img" aria-label="coverage: 80.35%"><title>coverage: 80.35%</title><filter id="blur"><feGaussianBlur stdDeviation="16"/></filter><linearGradient id="s" x2="0" y2="100%"><stop offset="0" stop-color="#bbb" stop-opacity=".1"/><stop offset="1" stop-opacity=".1"/></linearGradient><clipPath id="r"><rect width="114" height="20" rx="3"/></clipPath><g clip-path="url(#r)"><rect width="61" height="20" fill="#555"/><rect x="61" width="53" height="20" fill="#67ac09"/><rect width="114" height="20" fill="url(#s)"/></g><g fill="#fff" text-anchor="middle" font-family="Verdana,Geneva,DejaVu Sans,sans-serif" text-rendering="geometricPrecision" font-size="110"><g transform="scale(.1)"><g aria-hidden="true" fill="#010101"><text x="315" y="150" fill-opacity=".8" filter="url(#blur)" textLength="510">coverage</text><text x="315" y="150" fill-opacity=".3" textLength="510">coverage</text></g><text x="315" y="140" textLength="510">coverage</text></g><g transform="scale(.1)"><g aria-hidden="true" fill="#010101"><text x="865" y="150" fill-opacity=".8" filter="url(#blur)" textLength="430">80.35%</text><text x="865" y="150" fill-opacity=".3" textLength="430">80.35%</text></g><text x="865" y="140" textLength="430">80.35%</text></g></g></svg>
+<svg xmlns="http://www.w3.org/2000/svg" width="114" height="20" role="img" aria-label="coverage: 79.72%"><title>coverage: 79.72%</title><filter id="blur"><feGaussianBlur stdDeviation="16"/></filter><linearGradient id="s" x2="0" y2="100%"><stop offset="0" stop-color="#bbb" stop-opacity=".1"/><stop offset="1" stop-opacity=".1"/></linearGradient><clipPath id="r"><rect width="114" height="20" rx="3"/></clipPath><g clip-path="url(#r)"><rect width="61" height="20" fill="#555"/><rect x="61" width="53" height="20" fill="#67ac09"/><rect width="114" height="20" fill="url(#s)"/></g><g fill="#fff" text-anchor="middle" font-family="Verdana,Geneva,DejaVu Sans,sans-serif" text-rendering="geometricPrecision" font-size="110"><g transform="scale(.1)"><g aria-hidden="true" fill="#010101"><text x="315" y="150" fill-opacity=".8" filter="url(#blur)" textLength="510">coverage</text><text x="315" y="150" fill-opacity=".3" textLength="510">coverage</text></g><text x="315" y="140" textLength="510">coverage</text></g><g transform="scale(.1)"><g aria-hidden="true" fill="#010101"><text x="865" y="150" fill-opacity=".8" filter="url(#blur)" textLength="430">79.72%</text><text x="865" y="150" fill-opacity=".3" textLength="430">79.72%</text></g><text x="865" y="140" textLength="430">79.72%</text></g></g></svg>

{kolega_code-0.14.0 → kolega_code-0.15.0}/docs/src/content/docs/concepts/tools.md

@@ -18,6 +18,9 @@ Read, search, and edit your project:
 - `read_entire_file`, `read_file_section` — read file contents.
 - `find_files_by_pattern` — glob-based file search.
 - `search_codebase` — search the codebase by regular expression (ripgrep/grep), e.g. `foo|bar`.
+- `lsp` — read diagnostics, symbols, definitions,
+  references, hover text, call hierarchy, and code action metadata from
+  [configured language servers](../../configuration/lsp/).
 - Create and edit files — create new files and apply precise edits.
 
 ### Terminal

kolega_code-0.15.0/docs/src/content/docs/configuration/lsp.md

@@ -0,0 +1,121 @@
+---
+title: Language servers
+description: Configure LSP diagnostics, code intelligence, and trusted LSP edits.
+---
+
+Kolega Code can start local Language Server Protocol (LSP) servers for detected
+project languages. LSP is enabled by default and is used for:
+
+- diagnostics after `edit`, `multi_edit`, and `write`;
+- the generic read-only `lsp` tool;
+- the permission-gated `lsp_edit` tool;
+- `/lsp` and the Settings tab status display.
+
+The `lsp` tool remains read-only. Mutating LSP operations are exposed through
+`lsp_edit`, which is treated like `edit`, `multi_edit`, and `write` for
+permission prompts. Kolega Code does not expose raw arbitrary LSP requests.
+
+## Tool operations
+
+The generic `lsp` tool supports:
+
+- `diagnostics` for file errors, warnings, and hints;
+- `definition`, `type_definition`, `implementation`, `references`, and `hover`;
+- `document_symbols` and `workspace_symbols`;
+- `call_hierarchy` for incoming and outgoing calls;
+- `code_actions` to list available quick fixes and refactors without applying them;
+- `status`, `capabilities`, and `reload`.
+
+Position operations use `path`, a 1-based `line`, and a `symbol` on that line.
+Use `symbol#2` to select the second occurrence of a symbol on the same line.
+`code_actions` includes an `action_id` for each action so a later `lsp_edit`
+call can apply the exact server-provided action.
+
+The mutating `lsp_edit` tool supports:
+
+- `rename` for symbol rename;
+- `rename_file` for file moves with `workspace/willRenameFiles` and
+  `workspace/didRenameFiles` notifications;
+- `format_document` and `format_range`;
+- `apply_code_action` by `action_id`, title/query, or index.
+
+`lsp_edit` applies server-provided `WorkspaceEdit` payloads only. If the server
+does not advertise or return an edit for an operation, Kolega Code reports that
+the operation is unsupported instead of constructing a manual fallback edit.
+Pass `apply: false` to preview the LSP edit without writing files.
+
+## Project configuration
+
+Create `<project>/.kolega/lsp.json` to override LSP behavior for a repository.
+Project LSP config is trusted input because custom servers can execute local
+binaries. Kolega Code ignores this file unless the session is started with
+`--trust-lsp` or the project is listed in trusted LSP projects in user settings.
+
+```json
+{
+  "enabled": true,
+  "auto_diagnostics_on_edit": true,
+  "max_diagnostics": 20,
+  "auto_fallback": true,
+  "prompt_on_missing": true,
+  "disabled_languages": [],
+  "preferences": {
+    "python": "pyright"
+  },
+  "diagnostic_servers": ["ruff-lsp"],
+  "initialization_options": {
+    "pyright": {
+      "python": {
+        "analysis": {
+          "typeCheckingMode": "basic"
+        }
+      }
+    }
+  },
+  "workspace_configuration": {
+    "pyright": {
+      "python": {
+        "analysis": {
+          "typeCheckingMode": "basic"
+        }
+      }
+    }
+  },
+  "servers": {
+    "custom-python-lsp": {
+      "bin": "custom-lsp",
+      "args": ["--stdio"],
+      "languages": ["python"],
+      "env": {
+        "CUSTOM_LSP_MODE": "project"
+      }
+    }
+  }
+}
+```
+
+The project file cannot change the user's global LSP kill switch: `enabled` is
+accepted in user settings, but ignored from `.kolega/lsp.json`.
+
+`preferences` chooses the preferred server for a language. If
+`auto_fallback` is `false`, Kolega Code checks only the preferred server before
+reporting the language as missing. If it is `true`, installed alternatives can be
+used automatically.
+
+`diagnostic_servers` starts additional servers for diagnostics alongside the
+primary server. This is useful for setups such as Pyright plus Ruff.
+
+`initialization_options` is sent in the LSP `initialize` request. Some servers
+also ask for `workspace/configuration`; use `workspace_configuration` for those
+responses, keyed first by server name and then by requested section.
+
+## Status and troubleshooting
+
+Run `/lsp` in the TUI to see detected languages, missing servers, active
+sessions, and recent diagnostic counts. The Settings tab has a global LSP toggle;
+changing it requires restarting the agent session.
+
+If a server is missing, `/lsp` shows the server name and install guidance from
+Kolega Code's bundled registry. You can suppress a language with
+`disabled_languages` or point it at a custom server with `servers` and
+`preferences`.

{kolega_code-0.14.0 → kolega_code-0.15.0}/docs/src/content/docs/gigacode.mdx

@@ -34,6 +34,9 @@ can:
   modules, services, issues, or review targets.
 - **Run judge panels** — ask several agents to evaluate a change or plan from
   different angles, then reconcile disagreements.
+- **Compose workflow shapes** — use sweeps, matrices, funnels, pipelines, triage
+  loops, and adversarial reviews when the task needs more structure than simple
+  parallelism.
 - **Preserve artifacts** — save the workflow script, result files, transcripts, raw
   JSONL, and resume journal so the run is inspectable after the fact.
 
@@ -99,6 +102,13 @@ findings = await parallel([
 return findings
 ```
 
+The stage names in a workflow are not fixed runtime types. The agent composes
+shapes from the task: a broad audit might map surfaces, sweep reviewers across
+them, verify findings adversarially, and synthesize; a planning task might funnel
+unknowns through research and option comparison; a debugging task might cluster
+failures, investigate root causes, test targeted fixes, and run a regression
+review. All of those shapes are built from the same primitives above.
+
 <Aside type="note">
 Workflows are deterministic and **resumable**. Each run is saved under Kolega
 Code's state directory (`workflows/<run-id>/`) with its script, full result files,

{kolega_code-0.14.0 → kolega_code-0.15.0}/kolega_code/__init__.py

@@ -151,4 +151,4 @@ __all__ = [
     "parse_git_status_output",
 ]
 
-__version__ = "0.14.0"
+__version__ = "0.15.0"

{kolega_code-0.14.0 → kolega_code-0.15.0}/kolega_code/agent/__init__.py

@@ -44,4 +44,4 @@ __all__ = [
     "ToolExtension",
 ]
 
-__version__ = "0.14.0"
+__version__ = "0.15.0"

{kolega_code-0.14.0 → kolega_code-0.15.0}/kolega_code/agent/orchestration/guide.py

@@ -88,6 +88,23 @@ without running the script. Required keys: `name`, `description`. Optional `phas
 
 ### Quality patterns to compose
 
+- Surface map: before a broad run, map the real work surface — files, modules,
+  services, risks, unknowns, owner boundaries, or test targets — then fan out over
+  that map instead of guessing stages from the user's wording.
+- Shard-and-sweep: split a large surface by package/module/service/concern, send
+  focused agents across the shards, then deduplicate and synthesize.
+- Cross-cut matrix: review the same target set across dimensions such as correctness,
+  security, performance, UX, compatibility, test coverage, and migration risk.
+- Research funnel: map unknowns → research in parallel → compare options → produce
+  a decision-ready plan. In plan mode this is the main workflow shape.
+- Hypothesis tournament: generate several plausible explanations or approaches,
+  gather evidence for each, score them with parallel judges, then choose.
+- Implementation pipeline: map disjoint workstreams → implement with coder agents →
+  verify each stream → integrate and run final checks yourself.
+- Migration factory: inventory affected areas → classify risk → batch independent
+  edits → run targeted checks → compatibility review.
+- Failure triage loop: collect failures → cluster by likely root cause → investigate
+  clusters → fix isolated causes → rerun targeted tests.
 - Adversarial verify: for each finding, spawn N skeptics prompted to REFUTE it; keep it
   only if a majority fail to refute. Prevents plausible-but-wrong findings surviving.
 - Loop-until-dry: for unknown-size discovery, keep spawning finders until K consecutive
@@ -96,6 +113,8 @@ without running the script. Required keys: `name`, `description`. Optional `phas
   judges, synthesize from the winner.
 - Loop-until-budget: `while budget.total and budget.remaining() > 50_000: ...` to scale
   depth to the token ceiling.
+- Synthesis gate: after any broad fan-out, merge duplicates, rank confidence, identify
+  unresolved gaps, and decide whether another loop is worth the remaining budget.
 
 ### Implementing a plan (build mode)
 

{kolega_code-0.14.0 → kolega_code-0.15.0}/kolega_code/agent/tool_backend/edit_tool.py

@@ -1,10 +1,15 @@
+from __future__ import annotations
+
 import re
 from dataclasses import dataclass
-from typing import Callable
+from typing import TYPE_CHECKING, Callable, Optional
 
 from .base_tool import BaseTool
 from .edit_preview import build_diff_preview, build_head_preview
 
+if TYPE_CHECKING:
+    from kolega_code.services.lsp import LspManager
+
 
 _BLOCK_PATTERN = re.compile(r"<<<<<<< SEARCH\r?\n(.*?)\r?\n=======\r?\n(.*?)\r?\n>>>>>>> REPLACE", re.DOTALL)
 _SMART_PUNCTUATION_TRANSLATION = str.maketrans(
@@ -38,6 +43,10 @@ class ResolvedReplacement:
 
 
 class EditTool(BaseTool):
+    def __init__(self, *args, lsp_manager: Optional["LspManager"] = None, **kwargs) -> None:
+        super().__init__(*args, **kwargs)
+        self._lsp_manager = lsp_manager
+
     async def edit(self, path: str, block: str) -> str:
         """
         Edit a file using a single search and replace block.
@@ -54,7 +63,11 @@ class EditTool(BaseTool):
             error_msg = "Multiple search and replace blocks provided. Use multi_edit for multiple blocks."
             await self.log_error(error_msg, sender=self.caller.agent_name)
             raise ValueError(error_msg)
-        return await self._edit_blocks(path, parsed_blocks, tool_name="edit")
+        result = await self._edit_blocks(path, parsed_blocks, tool_name="edit")
+        lsp_diags = await self._maybe_append_lsp_diagnostics(path)
+        if lsp_diags:
+            result += lsp_diags
+        return result
 
     async def multi_edit(self, path: str, blocks: str) -> str:
         """
@@ -71,7 +84,11 @@ class EditTool(BaseTool):
             A short summary of the update
         """
         parsed_blocks = self._parse_blocks(blocks)
-        return await self._edit_blocks(path, parsed_blocks, tool_name="multi_edit")
+        result = await self._edit_blocks(path, parsed_blocks, tool_name="multi_edit")
+        lsp_diags = await self._maybe_append_lsp_diagnostics(path)
+        if lsp_diags:
+            result += lsp_diags
+        return result
 
     async def write(self, path: str, content: str) -> str:
         """
@@ -119,7 +136,11 @@ class EditTool(BaseTool):
                 tool_call_id=getattr(self.caller, "current_tool_execution_id", None),
                 tool_name="write",
             )
-            return f"Wrote {path}"
+            result = f"Wrote {path}"
+            lsp_diags = await self._maybe_append_lsp_diagnostics(path)
+            if lsp_diags:
+                result += lsp_diags
+            return result
         except PermissionError:
             error_msg = f"Permission denied when writing to file: {path}"
             await self.log_error(error_msg, sender=self.caller.agent_name)
@@ -379,3 +400,35 @@ class EditTool(BaseTool):
         if target == "\n":
             return text
         return text.replace("\n", target)
+
+    async def _maybe_append_lsp_diagnostics(self, path: str) -> str:
+        """Query LSP diagnostics for *path* and format them for appending to the tool result.
+
+        Returns an empty string if LSP is unavailable, disabled, or produces no diagnostics.
+        """
+        if self._lsp_manager is None or not self._lsp_manager.enabled:
+            return ""
+
+        if not self._lsp_manager._config.auto_diagnostics_on_edit:
+            return ""
+
+        if not self._lsp_manager._initialized:
+            await self._lsp_manager.initialize()
+
+        server_name = self._lsp_manager.server_for_path(path)
+        if server_name is None:
+            return ""
+
+        try:
+            diagnostics = await self._lsp_manager.get_fresh_diagnostics(path)
+        except Exception:
+            return ""
+
+        if not diagnostics:
+            return ""
+
+        from kolega_code.services.lsp import format_diagnostics
+
+        # Blank-line separator so the diagnostics block reads as a distinct section
+        # after the result line (e.g. "Edited foo.py\n\nLSP diagnostics (...)").
+        return "\n\n" + format_diagnostics(diagnostics, path, source=server_name)