perf: build system prompt once per turn, skip tools on force-text (#583)

* perf: build system prompt once per turn, skip tools on force-text, fix nudge role (#565)

Three fixes to agentic loop prompt handling:

1. Build system prompt once per turn instead of every tool iteration.
   `build_system_prompt_with_tools` is now pub; callers pass the result
   via `ReasoningContext::system_prompt` to avoid rebuilding ~1,500 tokens
   per iteration.

2. Skip `## Available Tools` section when `force_text = true`. The
   dispatcher passes a no-tools prompt variant on the final iteration,
   saving ~460 tokens and removing misleading instructions.

3. Change nudge message from `Role::System` to `Role::User`. A second
   system message mid-conversation is unsupported by most providers.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

* fix: revert nudge role change to keep ChatMessage::system

Copilot review correctly identified that using Role::User for the nudge
breaks compact_messages_for_retry, which uses rposition for Role::User
to find the last real user message. Role::Assistant would cause
back-to-back assistant messages. Since no production issues were reported
with the original system role, revert to ChatMessage::system.

[skip-regression-check]

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

* fix: address PR review — omit tool guidance when tools empty, rename shadowed var

- Conditionalize "Call tools…" guidelines and "## Tool Call Style" section
  in the system prompt so they are only included when tools are non-empty.
  Previously the force-text (no-tools) prompt still contained misleading
  tool-calling instructions. (Copilot review comment)

- Rename `system_prompt` → `cached_prompt` in dispatcher to avoid shadowing
  the earlier workspace identity `system_prompt` variable. (Copilot review)

- Add regression tests: `test_system_prompt_with_tools_contains_tool_guidance`
  and extended assertions in `test_system_prompt_without_tools_omits_tools_section`.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <[email protected]>
Co-authored-by: [email protected] <[email protected]>
This commit is contained in:
Henry Park
2026-03-07 09:15:00 +00:00
committed by GitHub
co-authored by Claude Opus 4.6 [email protected] <[email protected]>
parent 424a0366a9
commit 30790439ee
14 changed files with 400 additions and 321 deletions
+43 -1
View File
@@ -1,4 +1,4 @@
#![cfg(all(feature = "postgres", feature = "integration"))]
#![cfg(feature = "postgres")]
//! Integration tests for the workspace module.
//!
//! Requires a running PostgreSQL with pgvector extension.
@@ -21,6 +21,18 @@ fn get_pool() -> deadpool_postgres::Pool {
.expect("Failed to create pool")
}
/// Try to get a connection, returning None if Postgres is unreachable.
/// Tests call this to skip gracefully in CI where no database is available.
async fn try_connect(pool: &deadpool_postgres::Pool) -> Option<()> {
match pool.get().await {
Ok(_) => Some(()),
Err(e) => {
eprintln!("skipping: database unavailable ({e})");
None
}
}
}
async fn cleanup_user(pool: &deadpool_postgres::Pool, user_id: &str) {
let conn = pool.get().await.expect("Failed to get connection");
conn.execute(
@@ -34,6 +46,9 @@ async fn cleanup_user(pool: &deadpool_postgres::Pool, user_id: &str) {
#[tokio::test]
async fn test_workspace_write_and_read() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_write_read";
cleanup_user(&pool, user_id).await;
@@ -59,6 +74,9 @@ async fn test_workspace_write_and_read() {
#[tokio::test]
async fn test_workspace_append() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_append";
cleanup_user(&pool, user_id).await;
@@ -86,6 +104,9 @@ async fn test_workspace_append() {
#[tokio::test]
async fn test_workspace_nested_paths() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_nested";
cleanup_user(&pool, user_id).await;
@@ -131,6 +152,9 @@ async fn test_workspace_nested_paths() {
#[tokio::test]
async fn test_workspace_delete() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_delete";
cleanup_user(&pool, user_id).await;
@@ -155,6 +179,9 @@ async fn test_workspace_delete() {
#[tokio::test]
async fn test_workspace_memory_operations() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_memory_ops";
cleanup_user(&pool, user_id).await;
@@ -183,6 +210,9 @@ async fn test_workspace_memory_operations() {
#[tokio::test]
async fn test_workspace_daily_log() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_daily_log";
cleanup_user(&pool, user_id).await;
@@ -209,6 +239,9 @@ async fn test_workspace_daily_log() {
#[tokio::test]
async fn test_workspace_fts_search() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_fts_search";
cleanup_user(&pool, user_id).await;
@@ -267,6 +300,9 @@ async fn test_workspace_fts_search() {
#[tokio::test]
async fn test_workspace_hybrid_search_with_mock_embeddings() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_hybrid_search";
cleanup_user(&pool, user_id).await;
@@ -306,6 +342,9 @@ async fn test_workspace_hybrid_search_with_mock_embeddings() {
#[tokio::test]
async fn test_workspace_list_all() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_list_all";
cleanup_user(&pool, user_id).await;
@@ -331,6 +370,9 @@ async fn test_workspace_list_all() {
#[tokio::test]
async fn test_workspace_system_prompt() {
let pool = get_pool();
if try_connect(&pool).await.is_none() {
return;
}
let user_id = "test_system_prompt";
cleanup_user(&pool, user_id).await;