Conversation
The GitHub Release was titled "ralph v0.1.0" when it should be "scaffold v0.1.0" since scaffold is the top-level product. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
libagent.a now includes all dependency libraries (curl, mbedtls, cJSON, sqlite, uuid, python, zlib, pdfio) so consumers only need `libagent.a -lm -lpthread`. Uses GNU ar MRI scripts to merge both x86_64 and aarch64 archives independently. Release artifacts renamed to libagent.x86_64.a and libagent.aarch64.a. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Route all tool display through display_streaming_tool_start (now with id param), display_streaming_tool_delta, and display_streaming_tool_result instead of mixing direct printf, status_line_*, and log_tool_execution_improved calls. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Read-only query returning messages between two agents with cursor-based pagination. Does not mark messages as read (unlike message_receive_direct). Needed by ralphbot's GET /api/v1/messages endpoint. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…message ordering - Add async_executor_continue() wrapping session_continue() in a background thread, mirroring how async_executor_start() wraps session_process_message() - Make effects field optional in goap_create_actions (defaults to []), matching the existing preconditions default behavior - Add mkdir -p guard for fat archive output directory in lib.mk - Fix flaky test_message_store: message_list_between() ordered by created_at which was non-deterministic when messages shared the same millisecond timestamp; switched to rowid ordering for stable insertion-order results - Add tests for async_executor_continue() and optional effects behavior - Update test_create_actions_all_fail to test missing description (not effects) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Enables installing py3-none-any wheels from PyPI into ~/.local/{app}/site-packages/.
Packages load automatically via sitecustomize.py path injection and persist across
sessions. Includes _ralph_sys C extension for app path discovery, _ralph_http.download()
for binary-safe file downloads, and zip-slip protection in wheel extraction.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…load() 14 new tests covering: download() function existence and error handling, _ralph_sys module import and return types, _version_key pre-release ordering, _find_best_wheel wheel selection/rejection, _safe_extractall zip-slip protection, and pip_list empty directory handling. Also fixes _version_key bug where stable releases sorted below pre-releases due to missing release sentinel in tuple comparison. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Spinner now consults tool_extension_get_match_arg() to extract the display parameter for extension tools, so pip_install shows "pip_install (six)" instead of bare "pip_install". This makes the existing Match: docstring directive actually drive the UI display rather than relying solely on hardcoded parameter name chains. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- goal_store_has_active_goals: query whether any ACTIVE goals exist - goap_create_goal: new `persistent` param promotes goal to ACTIVE status - goap_update_world_state: new `summary` param for context recovery - 4 new goal_store tests for has_active_goals Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…t/summary - Check return values of goal_store_update_status and goal_store_update_summary - Add status field to goap_create_goal response - Use cJSON for summary extraction instead of redundant extract_string_param - Add tool-level tests for persistent goal creation and world state summary - Update ARCHITECTURE.md and CODE_OVERVIEW.md with new features and test counts Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Returns the latest N messages in ASC order with has_more flag and before_id cursor for backward pagination. Uses limit+1 fetch strategy to detect additional older messages without a separate COUNT query. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove the redundant ralph binary and make scaffold the sole output. The ralph binary was a strict subset of scaffold, so maintaining both added complexity without benefit. Key changes: - Move src/ralph/tools/ to src/tools/, delete src/ralph/ entirely - Remove ralph target from build system (single TARGET, OBJECTS, embed-python) - Remove all app_home_get_app_name() feature gating in lib/ — GOAP tools, orchestrator, /goals command, and scaffold prompt are now always active - Consolidate prompts: delete system.txt, rename SCAFFOLD_SYSTEM_PROMPT_TEXT to SYSTEM_PROMPT_TEXT - Change app_home default from "ralph" to "scaffold" - Fix executable_path.c fallback from "./ralph" to "./scaffold" - Fix GITHUB_REPO from "ralph" to "scaffold" (matches actual repo) - Rename protected config file from ralph.config.json to scaffold.config.json - Update CI docker image tag from ralph-dev to scaffold-dev - Update all tests, documentation, and copilot-instructions Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Create docs/ with 8 guides covering getting started, configuration, interactive mode, autonomous mode, memory, custom tools, MCP servers, CLI reference, and building from source. Update README with quick start section and documentation index linking to all guides. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Message ordering queries used ORDER BY created_at which is non-deterministic when messages share the same millisecond timestamp. Switch to ORDER BY rowid for guaranteed insertion order (matching message_list_between which already did this correctly). Build script now uses -j$(nproc) for parallel compilation, and CI uses build.sh instead of invoking make directly — single source of truth for the build command so issues surface locally, not just in CI. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- search_files: add context_lines param, matched_files list, total_matches_found counter, use finditer() for all matches per line, increase MAX_LINE_DISPLAY to 500 - read_file: return dict with line-numbered content, total_lines, range instead of plain string - list_dir: convert st_mtime to ISO timestamp strings - apply_delta: add optional expected field on replace/delete ops for stale-content detection, verify all expected fields before mutations - Add 4 integration tests covering all new features Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Split the monolithic system prompt into a stable base prefix and per-request dynamic context (todo state, mode, memories, context retrieval). This enables cache hits across requests within a session: - OpenAI: two system messages in messages array; first stays byte-identical across requests for automatic prefix caching - Anthropic: system field as JSON array of content blocks with cache_control ephemeral markers on the base block and last tool Rewrote build_json_payload_common to use cJSON throughout instead of manual snprintf JSON assembly. Empty todo lists no longer emit dynamic context, preserving cache stability. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Test plan
./scripts/build.sh./scripts/run_tests.sh--debugflag🤖 Generated with Claude Code