refactor(agent-tools): trim dead AgentChannel scaffolding to only doom_loop_hash

CHANGELOG.md

@@ -5,6 +5,19 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 
 ## [Unreleased]
+### Removed
+
+- `zeph-agent-tools`: removed the unused `AgentChannel` sealed trait and its supporting
+  `ChannelSinkError`/`ToolEventStart`/`ToolEventOutput`/`Sealed` types (`channel.rs`,
+  `error.rs`, `sealed.rs`), along with `ToolDispatchError` (`error.rs`) — dead scaffolding
+  from an abandoned dispatcher-extraction plan (issue #3516, closed) with zero implementors
+  or consumers anywhere in the workspace (#6480). The crate now exposes only `doom_loop_hash`,
+  its sole real production consumer (`zeph-core`'s tool tier loop). Dropped the now-unused
+  `futures`, `serde`, `serde_json`, `thiserror`, `tokio`, `tracing`, `zeph-common`, `zeph-llm`,
+  and `zeph-tools` dependencies along with the `sqlite`/`postgres` feature forwarding they
+  existed to support — `zeph-agent-tools` now has zero workspace or external dependencies.
+  Behavior of `doom_loop_hash` and its call site are unchanged.
+
 ### Changed
 
 - `zeph-skills`: `embed_skills_with_timeout` now tries `LlmProvider::embed_batch` first

Cargo.toml

@@ -160,7 +160,7 @@ zeph-acp = { path = "crates/zeph-acp", default-features = false, version = "0.22
 zeph-agent-context = { path = "crates/zeph-agent-context", default-features = false, version = "0.22.2" }
 zeph-agent-feedback = { path = "crates/zeph-agent-feedback", version = "0.22.2" }
 zeph-agent-persistence = { path = "crates/zeph-agent-persistence", default-features = false, version = "0.22.2" }
-zeph-agent-tools = { path = "crates/zeph-agent-tools", default-features = false, version = "0.22.2" }
+zeph-agent-tools = { path = "crates/zeph-agent-tools", version = "0.22.2" }
 zeph-bench = { path = "crates/zeph-bench", default-features = false, version = "0.22.2" }
 zeph-channels = { path = "crates/zeph-channels", default-features = false, version = "0.22.2" }
 zeph-commands = { path = "crates/zeph-commands", version = "0.22.2" }

crates/zeph-agent-context/src/lib.rs

@@ -18,7 +18,7 @@
 //! This crate provides [`service::ContextService`] — a stateless façade for all
 //! context-assembly operations that were previously implemented directly on `Agent<C>`
 //! in `zeph-core`. Extracting this logic means that editing context-assembly code does
-//! not trigger recompilation of the tool dispatcher (`zeph-agent-tools`) or the
+//! not trigger recompilation of the doom-loop detector (`zeph-agent-tools`) or the
 //! persistence layer (`zeph-agent-persistence`).
 //!
 //! # Architecture

crates/zeph-agent-tools/AGENTS.md

@@ -1,11 +1,8 @@
 # zeph-agent-tools Guide
 
-Tool-dispatch primitives consumed by the tool loop in `zeph-core`: the sealed `AgentChannel` trait, borrowed event carriers (`ToolEventStart`, `ToolEventOutput`), and doom-loop detection (`doom_loop_hash`).
+Doom-loop detection (`doom_loop_hash`) consumed by the tool loop in `zeph-core`.
 
 - Start with crate-local checks: `cargo build -p zeph-agent-tools`, `cargo nextest run -p zeph-agent-tools`, `cargo clippy -p zeph-agent-tools --all-targets -- -D warnings`.
-- Read `specs/006-tools/spec.md` before changing the dispatch contract or tool-result handling.
-- Architecture invariant: this crate MUST NOT depend on `zeph-core` or `zeph-channels`. `AgentChannel` is a minimal, sealed trait specifically to avoid the circular dependency that `zeph-core::channel::Channel` would create — never break this by adding those deps.
-- `AgentChannel` is sealed via the `Sealed` trait: external implementations are forbidden by design. `zeph-core` implements it through its local `AgentChannelView<'a, C>` adapter.
-- Crate status: Phase-2 scaffolding (issue #3516, closed). The `AgentChannel` trait and borrowed event carriers are complete, but no `zeph-core` adapter implements them and no `ToolDispatcher` extraction has landed or is in flight — that plan was abandoned, not deferred; re-opening it requires a new tracking issue. Per #6222/#6084, the crate now declares only the 3 `zeph-*` deps it actually uses (`zeph-common`, `zeph-llm`, `zeph-tools`) — do not re-add `zeph-agent-persistence`/`zeph-config`/`zeph-context`/`zeph-mcp`/`zeph-orchestration`/`zeph-sanitizer`/`zeph-skills` speculatively; add a dependency only alongside the code that actually needs it.
+- Architecture invariant: this crate MUST NOT depend on `zeph-core` or `zeph-channels`.
+- Crate status: previously carried a sealed `AgentChannel` dispatcher-extraction trait (issue #3516, closed) with zero implementors anywhere in the workspace; removed as dead code (issue #6480). That dispatcher-extraction plan is abandoned, not deferred — reviving it requires a new tracking issue. Add a dependency only alongside the code that actually needs it.
 - Doom-loop detection is agent-safety critical: any change to `doom_loop_hash` or its hashing inputs needs regression coverage so repeated tool calls are still detected.
-- LLM serialization gate: once tool dispatch / batch tool-result processing is extracted here, changes to those paths require a live session test with a real tool call before merge.

crates/zeph-agent-tools/Cargo.toml

@@ -9,27 +9,8 @@ homepage.workspace = true
 keywords.workspace = true
 categories.workspace = true
 publish.workspace = true
-description = "Agent tool dispatcher for Zeph: native tool loop, retry, confirmation, batch tool-result processing."
+description = "Doom-loop detection utilities for the Zeph agent tool dispatch loop."
 readme = "README.md"
 
-[dependencies]
-futures.workspace = true
-serde = { workspace = true, features = ["derive"] }
-serde_json.workspace = true
-thiserror.workspace = true
-tokio = { workspace = true, features = ["macros", "rt", "sync", "time"] }
-tracing.workspace = true
-zeph-common.workspace = true
-zeph-llm.workspace = true
-zeph-tools.workspace = true
-
-[features]
-# `zeph-tools` (and transitively `zeph-db`) requires a backend to compile;
-# default to `sqlite` so this crate builds in isolation, and expose `postgres` for
-# PostgreSQL deployments (#4956).
-default = ["sqlite"]
-sqlite = ["zeph-tools/sqlite"]
-postgres = ["zeph-tools/postgres"]
-
 [lints]
 workspace = true

crates/zeph-agent-tools/README.md

@@ -5,25 +5,16 @@
 [![License: MIT OR Apache-2.0](https://img.shields.io/badge/license-MIT%20OR%20Apache--2.0-blue.svg)](../../LICENSE)
 [![MSRV](https://img.shields.io/badge/MSRV-1.97-blue)](https://www.rust-lang.org)
 
-Agent tool dispatcher for Zeph: provides the `AgentChannel` sealed trait, borrowed event
-carriers, and doom-loop detection utilities used by the tool dispatch loop in `zeph-core`.
+Doom-loop detection utilities for the Zeph tool dispatch loop.
 
 ## Key types
 
 | Type / Function | Description |
 |-----------------|-------------|
-| `AgentChannel` | Sealed async sink trait the tool dispatcher uses to emit events to the user surface |
-| `ToolEventStart<'a>` | Borrowed payload for a tool-start event (zero-copy through the seam) |
-| `ToolEventOutput<'a>` | Borrowed payload for a tool-output event, including body and error flag |
-| `ChannelSinkError` | Concrete error returned by every `AgentChannel` method |
-| `ToolDispatchError` | Error enum covering LLM, tool, MCP, timeout, and channel failures |
 | `doom_loop_hash` | Hash message content with volatile tool IDs normalized out |
-| `Sealed` | Marker trait that prevents external `AgentChannel` implementations |
 
 ## Usage
 
-### Doom-loop detection
-
 ```rust
 use zeph_agent_tools::doom_loop_hash;
 
@@ -34,40 +25,14 @@ let h2 = doom_loop_hash("[tool_result: xyz789] same output");
 assert_eq!(h1, h2);
 ```
 
-### Implementing `AgentChannel` (zeph-core only)
-
-```rust,no_run
-// The trait is sealed — only zeph-core can implement it.
-// zeph-core provides: impl<C: Channel> AgentChannel for AgentChannelView<'_, C>
-//
-// Dispatcher usage (generic, no Box/dyn):
-async fn dispatch<Ch: zeph_agent_tools::AgentChannel>(ch: &mut Ch, text: &str) {
-    ch.send(text).await.ok();
-}
-```
-
 ## Architecture
 
-`zeph-agent-tools` does **not** depend on `zeph-core` or `zeph-channels`. It defines its own
-minimal `AgentChannel` trait (sealed via `Sealed`) which `zeph-core` implements through a local
-adapter type `AgentChannelView<'a, C>`. This avoids the circular dependency that would arise
-from referencing `zeph_core::channel::Channel` directly from inside the dispatcher crate.
-
-`zeph-core` depends on this crate; downstream channels (`zeph-channels`) do not.
-
-## Features
-
-| Feature | Default | Description |
-|---------|---------|-------------|
-| `sqlite` | on | SQLite backend, forwarded to `zeph-tools/sqlite` |
-| `postgres` | off | PostgreSQL backend, forwarded to `zeph-tools/postgres` |
-
-A storage backend must be selected for the crate to compile; `sqlite` is the default.
+`zeph-agent-tools` has no dependencies on other workspace crates. `zeph-core` calls
+`doom_loop_hash` from the tool dispatch loop to detect repeated tool-call cycles.
 
-> **Note:** This crate is Phase 2 scaffolding from issue #3516 (closed). The `AgentChannel`
-> trait and borrowed event carriers are complete, but no `zeph-core` adapter currently
-> implements them and no `ToolDispatcher` extraction has landed. Re-opening this work would
-> require a new tracking issue.
+> **Note:** This crate previously carried a sealed `AgentChannel` dispatcher-extraction trait
+> (issue #3516) with no implementors anywhere in the workspace; it was removed as dead code
+> (issue #6480). Reviving that dispatcher-extraction plan needs a fresh tracking issue.
 
 ## License