4 people 30790439ee 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]>

2026-03-07 09:15:00 +00:00

..

e2e

docs: add comprehensive subdirectory CLAUDE.md files and update root (#589)

2026-03-07 08:33:09 +00:00

fixtures/llm_traces

fix(llm): nudge LLM when it expresses tool intent without calling tools (#653)

2026-03-07 08:05:55 +00:00

support

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

2026-03-07 09:15:00 +00:00

test-pages

Feat/html to markdown #106 (#115)

2026-02-21 06:05:16 +00:00

config_round_trip.rs

Add automated QA: schema validator, CI matrix, Docker build, and P1 test coverage (#353)

2026-02-27 09:09:45 +04:00

e2e_advanced_traces.rs

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

2026-03-07 09:15:00 +00:00

e2e_builtin_tool_coverage.rs

test: add 29 E2E trace tests for issues #571-575 (#593)

2026-03-06 08:12:56 +00:00

e2e_metrics_test.rs

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

2026-03-07 09:15:00 +00:00

e2e_recorded_trace.rs

feat: merge http/web_fetch tools, add tool output stash for large responses (#578)

2026-03-06 00:49:10 +00:00

e2e_routine_heartbeat.rs

feat: Wire memory hygiene retention policy into heartbeat loop (#629)

2026-03-06 12:47:21 -08:00

e2e_safety_layer.rs

Trajectory benchmarks and e2e trace test rig (#553)

2026-03-05 09:13:09 +00:00

e2e_spot_checks.rs

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

2026-03-07 09:15:00 +00:00

e2e_status_events.rs

Trajectory benchmarks and e2e trace test rig (#553)

2026-03-05 09:13:09 +00:00

e2e_thread_scheduling.rs

test: add 29 E2E trace tests for issues #571-575 (#593)

2026-03-06 08:12:56 +00:00

e2e_tool_coverage.rs

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

2026-03-07 09:15:00 +00:00

e2e_trace_error_path.rs

Trajectory benchmarks and e2e trace test rig (#553)

2026-03-05 09:13:09 +00:00

e2e_trace_file_tools.rs

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

2026-03-07 09:15:00 +00:00

e2e_trace_memory.rs

Trajectory benchmarks and e2e trace test rig (#553)

2026-03-05 09:13:09 +00:00

e2e_worker_coverage.rs

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

2026-03-07 09:15:00 +00:00

e2e_workspace_coverage.rs

test: add 29 E2E trace tests for issues #571-575 (#593)

2026-03-06 08:12:56 +00:00

heartbeat_integration.rs

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

2026-03-07 09:15:00 +00:00

html_to_markdown.rs

feat: hot-activate WASM channels, channel-first prompts, unified artifact resolution (#297)

2026-02-22 08:09:56 +00:00

openai_compat_integration.rs

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

2026-03-07 09:15:00 +00:00

pairing_integration.rs

ci: Added CI/CD and release pipelines (#45)

2026-02-12 12:25:36 +01:00

provider_chaos.rs

feat: enable Anthropic prompt caching via automatic cache_control injection (#660)

2026-03-07 09:10:05 +00:00

support_unit_tests.rs

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

2026-03-07 09:15:00 +00:00

telegram_auth_integration.rs

fix(ci): fix three coverage workflow failures (#597)

2026-03-06 05:59:01 +00:00

tool_schema_validation.rs

test: add 26 tests for multi-thread safety, db CRUD, concurrency, errors (#442)

2026-03-06 04:39:04 +00:00

trace_format.rs

Trajectory benchmarks and e2e trace test rig (#553)

2026-03-05 09:13:09 +00:00

trace_llm_tests.rs

Trajectory benchmarks and e2e trace test rig (#553)

2026-03-05 09:13:09 +00:00

wasm_channel_integration.rs

feat(routines): deliver notifications to all installed channels (#398)

2026-02-27 12:01:32 -08:00

wit_compat.rs

feat: WASM extension versioning with WIT compat checks (#592)

2026-03-06 04:38:07 +00:00

workspace_integration.rs

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

2026-03-07 09:15:00 +00:00

ws_gateway_integration.rs

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

2026-03-07 09:15:00 +00:00