agentpack-cli 0.3.10__tar.gz → 0.3.11__tar.gz

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentpack-cli
-Version: 0.3.10
+Version: 0.3.11
 Summary: Local context engine for AI coding agents that ranks relevant files and builds task-focused context packs.
 License: MIT
 License-File: LICENSE
@@ -47,7 +47,7 @@ Description-Content-Type: text/markdown
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 [![CI](https://github.com/vishal2612200/agentpack/actions/workflows/ci.yml/badge.svg)](https://github.com/vishal2612200/agentpack/actions/workflows/ci.yml)
 
-> **Status: alpha (v0.3.10).** Works, tested, used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Public benchmark proof exists for the current suite, but broader repo coverage is still growing. API may change before 1.0.
+> **Status: alpha (v0.3.11).** Works, tested, used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Public benchmark proof exists for the current suite, but broader repo coverage is still growing. API may change before 1.0.
 >
 > **Platform note:** macOS, Linux, and Windows are supported. Windows support targets PowerShell plus Git for Windows. `cmd.exe` and bare Git setups are not a supported path yet.
 
@@ -65,6 +65,7 @@ Use AgentPack when a repo is too large to paste and you want faster, more consis
 - [Quality Bar](#quality-bar)
 - [Download Stats](#download-stats)
 - [Debugging Selection](#debugging-selection)
+- [Task Router](#task-router)
 - [Supported Integrations](#supported-integrations)
 - [Commands](#commands)
 - [Architecture](#architecture)
@@ -79,6 +80,7 @@ Use AgentPack when a repo is too large to paste and you want faster, more consis
 - **Local code intelligence**: extracts roles, domains, entrypoints, definitions, dependencies, env reads, side effects, and external systems using static analysis.
 - **Semantic repo map**: adds a compact module-level map before file context so agents orient faster.
 - **Freshness and deltas**: records task source, git state, snapshot hashes, selected-file deltas, stale-context warnings, MCP auto-refresh signals, and a machine-readable `agentpack:freshness` block in markdown fallback artifacts.
+- **Task router**: MCP and CLI surfaces route a task to relevant files, scoped rules, installed skills, suggested commands, and safety warnings without executing skills automatically.
 - **Agent integrations**: installs Claude Code, Cursor, Windsurf, Codex, Antigravity, VS Code tasks, git hooks, and MCP configuration.
 - **Local and measurable**: no API calls for scan, summarize, rank, pack, stats, or benchmark; quality is measured with expected-file evals.
 
@@ -292,6 +294,40 @@ agentpack guard --agent auto --repair-stale --refresh-context
 
 `guard` checks pack freshness, task freshness, repo snapshot freshness, and installed agent rules/hooks. With `--repair-stale --refresh-context`, it repairs stale AgentPack rule files and refreshes missing or stale context before returning success. `agentpack pack` also self-heals stale AgentPack rule blocks for the active agent, so older installs that still run `pack` get upgraded opportunistically.
 
+## Task Router
+
+AgentPack Router is the MCP-first path for agents that need a task map before loading full context. It returns:
+
+- files to read first
+- repo and tool rules to apply
+- installed skills to consider
+- commands to consider, never execute automatically
+- safety warnings for external side-effect skills
+- an agent-ready prompt block
+
+Use MCP when available:
+
+```text
+route_task("fix flaky payment webhook test")
+```
+
+Use CLI for inspection or scripting:
+
+```bash
+agentpack skills scan
+agentpack skills index
+agentpack route --task "fix flaky payment webhook test"
+agentpack route --task "fix flaky payment webhook test" --format json
+```
+
+Router reads skills and rules from `.claude/skills/`, `~/.claude/skills/`, `~/.codex/skills/`, `~/.agents/skills/`, `.agentpack/skills/`, `.cursor/rules/`, `AGENTS.md`, `CLAUDE.md`, and `GEMINI.md`. Rules are mandatory scoped instructions; skills are optional recommendations. The local `.agentpack/skills_index.json` stores metadata only and omits raw skill/rule bodies.
+
+Safety defaults:
+
+- skills are recommended, not executed
+- suggested commands are returned as strings with reasons
+- external side-effect skills, such as deploy or cloud mutation checklists, are warned and not selected unless explicitly allowed in config
+
 ## Before / After Agent Behavior
 
 Without AgentPack:
@@ -598,6 +634,9 @@ Command map:
 | `agentpack install` | Refresh or add an agent integration without changing project state |
 | `agentpack repair` | Restore missing or drifted integration files |
 | `agentpack pack` | Generate a ranked context pack for one task |
+| `agentpack route` | Route a task to files, rules, skills, commands, and safety warnings |
+| `agentpack skills scan` | Print discovered local/global skills and rules |
+| `agentpack skills index` | Write `.agentpack/skills_index.json` metadata for faster routing |
 | `agentpack watch` | Keep the context pack fresh while you work |
 | `agentpack doctor` | Audit hooks, agent files, CLI path, and repo health |
 | `agentpack explain` | Understand why a file was selected or omitted |
@@ -895,6 +934,32 @@ This keeps unrelated dirty files from consuming the whole context budget while p
 
 ---
 
+### `agentpack route`
+
+Route a task without writing context files. This is the CLI debug/admin surface for the same router used by MCP `route_task`.
+
+```bash
+agentpack route --task "fix flaky payment webhook test"
+agentpack route --task "fix flaky payment webhook test" --format json
+```
+
+Output includes relevant files, applied rules, recommended skills, suggested commands, safety warnings, and an agent prompt. It uses the existing AgentPack file ranker in memory and does not write `.agentpack/context.md`.
+
+---
+
+### `agentpack skills`
+
+Inspect or index installed skills and rule files.
+
+```bash
+agentpack skills scan
+agentpack skills index
+```
+
+`scan` prints discovered artifacts. `index` writes `.agentpack/skills_index.json` with metadata only; raw skill and rule bodies are omitted from the index.
+
+---
+
 ### `agentpack quickstart`
 
 Show the shortest useful path for the current repo.
@@ -986,6 +1051,9 @@ Register in Claude Code settings (`~/.claude/settings.json`):
 
 | Tool | Description |
 |---|---|
+| `route_task(task)` | Read-only task router. Returns relevant files, applied rules, recommended skills, suggested commands, safety warnings, and an agent prompt as JSON. |
+| `get_skills()` | Return discovered skill/rule inventory as JSON. |
+| `explain_route(task)` | Return route JSON with positive skill score reasons for debugging router choices. |
 | `start_task(task, mode, budget, max_tokens)` | Recommended MCP-first entry point. Writes `.agentpack/task.md`, generates a ranked pack, and returns packed markdown. |
 | `pack_context(task, mode, budget, max_tokens)` | Generate a ranked context pack. If `task` is provided, writes it to `.agentpack/task.md`; if omitted, reads `task.md` or infers from git. |
 | `get_context()` | Return the latest pack. If `.agentpack/task.md` or the repo snapshot differs from the packed metadata, it auto-refreshes before returning; otherwise it prepends a freshness header. |

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/README.md

@@ -8,7 +8,7 @@
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 [![CI](https://github.com/vishal2612200/agentpack/actions/workflows/ci.yml/badge.svg)](https://github.com/vishal2612200/agentpack/actions/workflows/ci.yml)
 
-> **Status: alpha (v0.3.10).** Works, tested, used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Public benchmark proof exists for the current suite, but broader repo coverage is still growing. API may change before 1.0.
+> **Status: alpha (v0.3.11).** Works, tested, used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Public benchmark proof exists for the current suite, but broader repo coverage is still growing. API may change before 1.0.
 >
 > **Platform note:** macOS, Linux, and Windows are supported. Windows support targets PowerShell plus Git for Windows. `cmd.exe` and bare Git setups are not a supported path yet.
 
@@ -26,6 +26,7 @@ Use AgentPack when a repo is too large to paste and you want faster, more consis
 - [Quality Bar](#quality-bar)
 - [Download Stats](#download-stats)
 - [Debugging Selection](#debugging-selection)
+- [Task Router](#task-router)
 - [Supported Integrations](#supported-integrations)
 - [Commands](#commands)
 - [Architecture](#architecture)
@@ -40,6 +41,7 @@ Use AgentPack when a repo is too large to paste and you want faster, more consis
 - **Local code intelligence**: extracts roles, domains, entrypoints, definitions, dependencies, env reads, side effects, and external systems using static analysis.
 - **Semantic repo map**: adds a compact module-level map before file context so agents orient faster.
 - **Freshness and deltas**: records task source, git state, snapshot hashes, selected-file deltas, stale-context warnings, MCP auto-refresh signals, and a machine-readable `agentpack:freshness` block in markdown fallback artifacts.
+- **Task router**: MCP and CLI surfaces route a task to relevant files, scoped rules, installed skills, suggested commands, and safety warnings without executing skills automatically.
 - **Agent integrations**: installs Claude Code, Cursor, Windsurf, Codex, Antigravity, VS Code tasks, git hooks, and MCP configuration.
 - **Local and measurable**: no API calls for scan, summarize, rank, pack, stats, or benchmark; quality is measured with expected-file evals.
 
@@ -253,6 +255,40 @@ agentpack guard --agent auto --repair-stale --refresh-context
 
 `guard` checks pack freshness, task freshness, repo snapshot freshness, and installed agent rules/hooks. With `--repair-stale --refresh-context`, it repairs stale AgentPack rule files and refreshes missing or stale context before returning success. `agentpack pack` also self-heals stale AgentPack rule blocks for the active agent, so older installs that still run `pack` get upgraded opportunistically.
 
+## Task Router
+
+AgentPack Router is the MCP-first path for agents that need a task map before loading full context. It returns:
+
+- files to read first
+- repo and tool rules to apply
+- installed skills to consider
+- commands to consider, never execute automatically
+- safety warnings for external side-effect skills
+- an agent-ready prompt block
+
+Use MCP when available:
+
+```text
+route_task("fix flaky payment webhook test")
+```
+
+Use CLI for inspection or scripting:
+
+```bash
+agentpack skills scan
+agentpack skills index
+agentpack route --task "fix flaky payment webhook test"
+agentpack route --task "fix flaky payment webhook test" --format json
+```
+
+Router reads skills and rules from `.claude/skills/`, `~/.claude/skills/`, `~/.codex/skills/`, `~/.agents/skills/`, `.agentpack/skills/`, `.cursor/rules/`, `AGENTS.md`, `CLAUDE.md`, and `GEMINI.md`. Rules are mandatory scoped instructions; skills are optional recommendations. The local `.agentpack/skills_index.json` stores metadata only and omits raw skill/rule bodies.
+
+Safety defaults:
+
+- skills are recommended, not executed
+- suggested commands are returned as strings with reasons
+- external side-effect skills, such as deploy or cloud mutation checklists, are warned and not selected unless explicitly allowed in config
+
 ## Before / After Agent Behavior
 
 Without AgentPack:
@@ -559,6 +595,9 @@ Command map:
 | `agentpack install` | Refresh or add an agent integration without changing project state |
 | `agentpack repair` | Restore missing or drifted integration files |
 | `agentpack pack` | Generate a ranked context pack for one task |
+| `agentpack route` | Route a task to files, rules, skills, commands, and safety warnings |
+| `agentpack skills scan` | Print discovered local/global skills and rules |
+| `agentpack skills index` | Write `.agentpack/skills_index.json` metadata for faster routing |
 | `agentpack watch` | Keep the context pack fresh while you work |
 | `agentpack doctor` | Audit hooks, agent files, CLI path, and repo health |
 | `agentpack explain` | Understand why a file was selected or omitted |
@@ -856,6 +895,32 @@ This keeps unrelated dirty files from consuming the whole context budget while p
 
 ---
 
+### `agentpack route`
+
+Route a task without writing context files. This is the CLI debug/admin surface for the same router used by MCP `route_task`.
+
+```bash
+agentpack route --task "fix flaky payment webhook test"
+agentpack route --task "fix flaky payment webhook test" --format json
+```
+
+Output includes relevant files, applied rules, recommended skills, suggested commands, safety warnings, and an agent prompt. It uses the existing AgentPack file ranker in memory and does not write `.agentpack/context.md`.
+
+---
+
+### `agentpack skills`
+
+Inspect or index installed skills and rule files.
+
+```bash
+agentpack skills scan
+agentpack skills index
+```
+
+`scan` prints discovered artifacts. `index` writes `.agentpack/skills_index.json` with metadata only; raw skill and rule bodies are omitted from the index.
+
+---
+
 ### `agentpack quickstart`
 
 Show the shortest useful path for the current repo.
@@ -947,6 +1012,9 @@ Register in Claude Code settings (`~/.claude/settings.json`):
 
 | Tool | Description |
 |---|---|
+| `route_task(task)` | Read-only task router. Returns relevant files, applied rules, recommended skills, suggested commands, safety warnings, and an agent prompt as JSON. |
+| `get_skills()` | Return discovered skill/rule inventory as JSON. |
+| `explain_route(task)` | Return route JSON with positive skill score reasons for debugging router choices. |
 | `start_task(task, mode, budget, max_tokens)` | Recommended MCP-first entry point. Writes `.agentpack/task.md`, generates a ranked pack, and returns packed markdown. |
 | `pack_context(task, mode, budget, max_tokens)` | Generate a ranked context pack. If `task` is provided, writes it to `.agentpack/task.md`; if omitted, reads `task.md` or infers from git. |
 | `get_context()` | Return the latest pack. If `.agentpack/task.md` or the repo snapshot differs from the packed metadata, it auto-refreshes before returning; otherwise it prepends a freshness header. |

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "agentpack-cli"
-version = "0.3.10"
+version = "0.3.11"
 description = "Local context engine for AI coding agents that ranks relevant files and builds task-focused context packs."
 readme = "README.md"
 requires-python = ">=3.10"

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/__init__.py

@@ -1,3 +1,3 @@
 """AgentPack — task-aware context packing for AI coding agents."""
 
-__version__ = "0.3.10"
+__version__ = "0.3.11"

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/cli.py

@@ -19,7 +19,9 @@ from agentpack.commands import (
     pack,
     quickstart,
     repair,
+    route,
     scan,
+    skills,
     stats,
     status,
     summarize,
@@ -56,6 +58,7 @@ for mod in [
     pack,
     install,
     repair,
+    route,
     migrate,
     monitor,
     explain,
@@ -69,6 +72,7 @@ for mod in [
     mcp_cmd,
     hook_cmd,
     quickstart,
+    skills,
 ]:
     mod.register(app)
 

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/commands/mcp_cmd.py

@@ -7,6 +7,6 @@ import typer
 def register(app: typer.Typer) -> None:
     @app.command("mcp")
     def mcp_server() -> None:
-        """Start the AgentPack MCP server (tools: pack_context, get_context, refresh)."""
+        """Start the AgentPack MCP server (tools: route_task, pack_context, get_context, refresh)."""
         from agentpack.mcp_server import serve
         serve()

agentpack_cli-0.3.11/src/agentpack/commands/route.py

@@ -0,0 +1,29 @@
+from __future__ import annotations
+
+import typer
+
+from agentpack.commands._shared import _root, console
+from agentpack.router.prompt_builder import render_plain
+from agentpack.router.service import RouteService
+
+
+def register(app: typer.Typer) -> None:
+    @app.command("route")
+    def route_task(
+        task: str = typer.Option(..., "--task", help="Developer task to route."),
+        output_format: str = typer.Option("plain", "--format", help="Output format: plain|json."),
+    ) -> None:
+        """Route a task to relevant files, rules, skills, and command suggestions."""
+        if output_format not in {"plain", "json"}:
+            console.print("[red]Invalid format. Use plain|json.[/]")
+            raise typer.Exit(1)
+        try:
+            result = RouteService().route_task(_root(), task)
+        except ValueError as exc:
+            console.print(f"[red]{exc}[/]")
+            raise typer.Exit(1) from exc
+
+        if output_format == "json":
+            typer.echo(result.model_dump_json(indent=2))
+        else:
+            console.print(render_plain(result))

agentpack_cli-0.3.11/src/agentpack/commands/skills.py

@@ -0,0 +1,46 @@
+from __future__ import annotations
+
+import typer
+from rich.table import Table
+
+from agentpack.commands._shared import _root, console
+from agentpack.core.config import load_config
+from agentpack.router.discovery import discover_inventory, write_inventory_index
+
+skills_app = typer.Typer(help="Inspect and index local agent skills and rules.")
+
+
+def register(app: typer.Typer) -> None:
+    app.add_typer(skills_app, name="skills")
+
+
+@skills_app.command("scan")
+def scan_skills() -> None:
+    """Print discovered skills and rules without writing an index."""
+    root = _root()
+    cfg = load_config(root)
+    inventory = discover_inventory(root, cfg.skills.paths)
+    console.print(f"Found {len(inventory.skills)} skills and {len(inventory.rules)} rules")
+
+    table = Table(show_header=True)
+    table.add_column("type")
+    table.add_column("name")
+    table.add_column("source")
+    table.add_column("description")
+    for skill in inventory.skills:
+        table.add_row("skill", skill.name, skill.path, skill.description[:80])
+    for rule in inventory.rules:
+        table.add_row("rule", rule.name, rule.path, rule.description[:80])
+    console.print(table)
+
+
+@skills_app.command("index")
+def index_skills() -> None:
+    """Write .agentpack/skills_index.json."""
+    root = _root()
+    cfg = load_config(root)
+    inventory = discover_inventory(root, cfg.skills.paths)
+    path = write_inventory_index(root, inventory)
+    console.print(
+        f"Indexed {len(inventory.skills)} skills and {len(inventory.rules)} rules at {path}"
+    )

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/core/config.py

@@ -42,6 +42,19 @@ class HooksConfig(BaseModel):
     blocking_task_refresh: bool = True
 
 
+class SkillsConfig(BaseModel):
+    paths: list[str] = Field(default_factory=lambda: [
+        ".claude/skills",
+        "~/.claude/skills",
+        "~/.codex/skills",
+        "~/.agents/skills",
+        ".agentpack/skills",
+        ".cursor/rules",
+    ])
+    max_selected: int = 3
+    allow_external_side_effects: bool = False
+
+
 class AgentConfig(BaseModel):
     output: str
     patch_claude_md: bool = False
@@ -89,6 +102,7 @@ class Config(BaseModel):
     context: ContextConfig = Field(default_factory=ContextConfig)
     summary: SummaryConfig = Field(default_factory=SummaryConfig)
     hooks: HooksConfig = Field(default_factory=HooksConfig)
+    skills: SkillsConfig = Field(default_factory=SkillsConfig)
     agents: AgentsConfig = Field(default_factory=AgentsConfig)
     scoring: ScoringWeights = Field(default_factory=ScoringWeights)
 
@@ -128,6 +142,12 @@ task_switch_min_terms = 1
 # Block once on task switches so the first prompt sees fresh top-file hints.
 blocking_task_refresh = true
 
+[skills]
+# Skill/rule sources used by `agentpack route` and MCP `route_task`.
+paths = [".claude/skills", "~/.claude/skills", "~/.codex/skills", "~/.agents/skills", ".agentpack/skills", ".cursor/rules"]
+max_selected = 3
+allow_external_side_effects = false
+
 [scoring]
 # Scoring weights — higher wins budget allocation.
 # Tune these to make agentpack favour your team's file layout.

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/installers/antigravity.py

@@ -14,9 +14,10 @@ skills:
 At the start of every coding task:
 1. Write a one-line task summary to `.agentpack/task.md` (overwrite the whole file).
 2. Run `agentpack guard --agent antigravity --repair-stale --refresh-context`. This is the executable pre-edit gate for non-MCP paths.
-3. Prefer AgentPack MCP if available. Call `agentpack_get_context()` for current context, or `agentpack_pack_context(task="<task>")` for a new task. MCP is the active path because it auto-refreshes stale task/repo context.
-4. If MCP is unavailable, read `.agent/skills/agentpack/SKILL.md`. Treat it as fallback; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
-5. Use files listed in context as starting points, but verify with actual code before editing.
+3. Prefer AgentPack MCP if available. MCP is the active path. Call `agentpack_route_task(task="<task>")` first to get files, rules, skills, commands, and safety warnings.
+4. Call `agentpack_pack_context(task="<task>")` only when full packed context is needed, or `agentpack_get_context()` for current context.
+5. If MCP is unavailable, read `.agent/skills/agentpack/SKILL.md`. Treat it as fallback; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
+6. Use files listed in context as starting points, but verify with actual code before editing.
 
 When the user switches to a different coding task, update `.agentpack/task.md`, then call MCP again or rerun the guard command before editing.
 <!-- agentpack:block:end -->"""

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/installers/claude.py

@@ -14,10 +14,10 @@ Prefer MCP over reading `.agentpack/context*.md` directly because MCP auto-refre
 and repo-snapshot context before returning.
 
 ```
-mcp__agentpack__get_context()
+mcp__agentpack__route_task(task="<what you're working on>")
 ```
 
-For a brand-new task, call:
+When full packed context is needed, call:
 
 ```
 mcp__agentpack__pack_context(task="<what you're working on>", budget=4000)
@@ -30,6 +30,7 @@ agentpack guard --agent claude --repair-stale --refresh-context
 ```
 
 Other tools:
+- `mcp__agentpack__route_task(task)` — files, rules, skills, commands, and safety warnings
 - `mcp__agentpack__explain_file(path)` — score breakdown + symbols for a file
 - `mcp__agentpack__get_related_files(path)` — import-graph neighbours
 - `mcp__agentpack__get_stats()` — token/saving stats for the latest pack

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/installers/codex.py

@@ -14,9 +14,10 @@ At the start of every coding task:
 
 1. Write a one-line task summary to `.agentpack/task.md` (overwrite the whole file).
 2. Run `agentpack guard --agent codex --repair-stale --refresh-context`. This is the executable pre-edit gate for non-MCP paths.
-3. Prefer AgentPack MCP if available. Call `agentpack_get_context()` for existing task context, or `agentpack_pack_context(task="<task>")` for a new task. MCP is the active path because it auto-refreshes stale task/repo context.
-4. If MCP is unavailable, read `.agentpack/context.md`. Treat it as a fallback artifact; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
-5. Use selected files as starting points, but verify with actual code before editing.
+3. Prefer AgentPack MCP if available. MCP is the active path. Call `agentpack_route_task(task="<task>")` first to get files, rules, skills, commands, and safety warnings.
+4. Call `agentpack_pack_context(task="<task>")` only when full packed context is needed, or `agentpack_get_context()` for existing task context.
+5. If MCP is unavailable, read `.agentpack/context.md`. Treat it as a fallback artifact; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
+6. Use selected files as starting points, but verify with actual code before editing.
 
 When the user switches to a different coding task, update `.agentpack/task.md`, then call MCP again or rerun the guard command before editing.
 <!-- agentpack:end -->"""

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/installers/cursor.py

@@ -11,9 +11,10 @@ _CURSOR_RULE = """\
 At the start of every coding task:
 1. Write a one-line task summary to `.agentpack/task.md` (overwrite the whole file).
 2. Run `agentpack guard --agent cursor --repair-stale --refresh-context`. This is the executable pre-edit gate for non-MCP paths.
-3. Prefer AgentPack MCP if available. Call `agentpack_get_context()` for current context, or `agentpack_pack_context(task="<task>")` for a new task. MCP is the active path because it auto-refreshes stale task/repo context.
-4. If MCP is unavailable, read `.agentpack/context.md`. Treat it as fallback; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
-5. Use files listed in context as starting points, but verify with actual code before editing.
+3. Prefer AgentPack MCP if available. MCP is the active path. Call `agentpack_route_task(task="<task>")` first to get files, rules, skills, commands, and safety warnings.
+4. Call `agentpack_pack_context(task="<task>")` only when full packed context is needed, or `agentpack_get_context()` for current context.
+5. If MCP is unavailable, read `.agentpack/context.md`. Treat it as fallback; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
+6. Use files listed in context as starting points, but verify with actual code before editing.
 When the user switches to a different coding task, update `.agentpack/task.md`, then call MCP again or rerun the guard command before editing.
 If context is missing: write `.agentpack/task.md`, then run `agentpack guard --agent cursor --repair-stale --refresh-context`.
 <!-- agentpack:rule:end -->"""
@@ -62,9 +63,10 @@ At the start of every coding task:
 
 1. Write a one-line task summary to `.agentpack/task.md` (overwrite the whole file).
 2. Run `agentpack guard --agent cursor --repair-stale --refresh-context`. This is the executable pre-edit gate for non-MCP paths.
-3. Prefer AgentPack MCP if available. Call `agentpack_get_context()` for current context, or `agentpack_pack_context(task="<task>")` for a new task. MCP is the active path because it auto-refreshes stale task/repo context.
-4. If MCP is unavailable, read `.agentpack/context.md`. Treat it as fallback; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
-5. Use files listed in context as starting points, but verify with actual code before editing.
+3. Prefer AgentPack MCP if available. MCP is the active path. Call `agentpack_route_task(task="<task>")` first to get files, rules, skills, commands, and safety warnings.
+4. Call `agentpack_pack_context(task="<task>")` only when full packed context is needed, or `agentpack_get_context()` for current context.
+5. If MCP is unavailable, read `.agentpack/context.md`. Treat it as fallback; if its `agentpack:freshness` block says `refresh_required: true` or the task does not match, rerun the guard command before using selected files.
+6. Use files listed in context as starting points, but verify with actual code before editing.
 
 When the user switches to a different coding task, update `.agentpack/task.md`, then call MCP again or rerun the guard command before editing.
 

{agentpack_cli-0.3.10 → agentpack_cli-0.3.11}/src/agentpack/mcp_server.py

@@ -16,6 +16,9 @@ Or register in Claude Code settings:
 Tools exposed:
     start_task          — write task.md and return a fresh context pack
     pack_context        — generate/refresh a context pack for a task
+    route_task          — read-only route: files + rules + skills + commands
+    get_skills          — read-only skill/rule inventory
+    explain_route       — read-only route with skill score reasons
     get_context         — read latest context pack; auto-refreshes when task.md changed
     refresh             — refresh using the current task.md
     explain_file        — show score breakdown + symbols for a specific file
@@ -420,6 +423,30 @@ def _get_delta_context_impl(root: Path, max_files: int = 12) -> str:
     return "\n".join(lines)
 
 
+def _route_task_impl(root: Path, task: str) -> str:
+    """Return read-only task route JSON; does not write task/context files."""
+    from agentpack.router.service import RouteService
+
+    result = RouteService().route_task(root, task)
+    return result.model_dump_json(indent=2)
+
+
+def _get_skills_impl(root: Path) -> str:
+    """Return discovered skill/rule inventory JSON."""
+    from agentpack.router.service import RouteService
+
+    inventory = RouteService().inventory(root)
+    return inventory.model_dump_json(indent=2)
+
+
+def _explain_route_impl(root: Path, task: str) -> str:
+    """Return task route JSON including all positive skill scores."""
+    from agentpack.router.service import RouteService
+
+    result = RouteService().explain_route(root, task)
+    return result.model_dump_json(indent=2)
+
+
 def serve() -> None:
     try:
         from mcp.server.fastmcp import FastMCP
@@ -468,6 +495,25 @@ def serve() -> None:
             max_tokens=max_tokens,
         )
 
+    @mcp.tool()
+    def route_task(task: str) -> str:
+        """Route a task to files, rules, skills, command suggestions, and safety warnings.
+
+        Read-only: does not write task.md or context files. Use pack_context when full
+        context content is needed.
+        """
+        return _route_task_impl(_repo_root(), task)
+
+    @mcp.tool()
+    def get_skills() -> str:
+        """Return the discovered Agentpack skill/rule inventory as JSON."""
+        return _get_skills_impl(_repo_root())
+
+    @mcp.tool()
+    def explain_route(task: str) -> str:
+        """Return a route_task-style JSON result with skill scoring reasons."""
+        return _explain_route_impl(_repo_root(), task)
+
     @mcp.tool()
     def get_context() -> str:
         """Return the latest context pack, auto-refreshing when task.md changed.

agentpack_cli-0.3.11/src/agentpack/router/__init__.py

@@ -0,0 +1,23 @@
+from agentpack.router.models import (
+    AppliedRule,
+    CommandSuggestion,
+    RouteExplanation,
+    RouteResult,
+    SelectedSkill,
+    SkillArtifact,
+    SkillInventory,
+    RuleArtifact,
+)
+from agentpack.router.service import RouteService
+
+__all__ = [
+    "AppliedRule",
+    "CommandSuggestion",
+    "RouteExplanation",
+    "RouteResult",
+    "RouteService",
+    "SelectedSkill",
+    "SkillArtifact",
+    "SkillInventory",
+    "RuleArtifact",
+]