Merged
Conversation
🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
andreasronge
force-pushed
the
test/ci-workflow-check
branch
from
0d232c9 to
ae5ad46
Compare
andreasronge
added a commit
that referenced
this pull request
Dec 1, 2025
…x execution (#10) * feat: implement Phase 1 core interpreter with JSON parsing and sandbox execution Implements the foundational Phase 1 of PtcRunner including: - **Parser**: JSON string and map parsing with error handling - **Validator**: DSL schema validation for all Phase 1 operations - **Context**: Variable bindings management - **Operations**: Core data, control flow, collection, comparison, and aggregation operations - Data: literal, load, var - Control: pipe - Collections: filter, map, select - Comparison: eq - Aggregations: sum, count - **Interpreter**: AST evaluation with operation dispatch - **Sandbox**: Isolated BEAM process execution with timeout and resource monitoring All programs execute in isolated processes with configurable timeouts and memory limits, returning execution metrics (duration_ms, memory_bytes). Includes comprehensive test coverage for all operations, error cases, and edge cases. Fixes #7 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: resolve PR review issues for validation and sandbox - Fix validate_list/1: Use Enum.reduce_while instead of Enum.find to properly validate all nested operations (Issue #1) - Fix sandbox memory limits: Pass max_heap option to Process.spawn to enforce memory limits (Issue #2) - Add test for nested validation errors: Verify validation catches unknown operations inside pipe steps (Issue #3) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: resolve PR review issues for timeout and memory errors - Fix timeout error format to match architecture.md: return {:error, {:timeout, timeout_value}} instead of {:error, :timeout} - Fix memory exceeded error detection: distinguish between memory killed (:killed reason) and timeout kills - Update sandbox.ex:72-76 to handle :killed reason as memory exceeded with {:error, {:memory_exceeded, max_heap * 8}} - Update sandbox.ex:82 to return {:error, {:timeout, timeout}} for explicit timeout kills - Update test case to expect new timeout error format {:timeout, 0} instead of :timeout 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * chore: update PM status - PR #10 Phase 1 implementation --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>
This was referenced Dec 2, 2025
4 tasks
This was referenced Dec 9, 2025
This was referenced Dec 30, 2025
claude bot
pushed a commit
that referenced
this pull request
Dec 31, 2025
- Capture fail tool arguments in error message instead of generic "Explicit fail called" - Line 233: Replace generic message with `inspect(lisp_step.return)` to preserve failure reason - Fixes PR review issue #2: failure data passed to fail tool is now preserved - Add comment linking to issue #374 for future system prompt enhancements - Line 313-314: Document that system prompt is intentionally minimal - Reference existing issue #374 for context-aware prompts, tool documentation, etc. - Fixes PR review issue #4: clarifies intentional design choice 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
andreasronge
pushed a commit
that referenced
this pull request
Dec 31, 2025
* feat: implement Loop.run/2 with response parsing Implement the core agentic loop that manages LLM↔tool cycles. The loop calls the LLM, parses PTC-Lisp from the response, executes it, and continues until return/fail is called or max_turns is exceeded. **Key Features:** - Multi-turn execution with LLM feedback loops - PTC-Lisp extraction from code blocks (clojure/lisp) or raw s-expressions - System tools (return/fail) for loop termination - Message history accumulation across turns - Trace entry generation per turn - Error feedback to LLM for recovery - Context merging between turns **Implementation Details:** - Detects explicit return/fail calls to terminate loop - Feeds execution results back to LLM for multi-turn workflows - Handles max_turns exceeded with appropriate error - Builds llm_input with system prompt, messages, turn number, and tool names - Makes previous turn errors available as ctx/fail **Testing:** - Comprehensive unit tests for all loop scenarios - Response parsing tests for different code block formats - Multi-turn execution tests - Error handling and recovery tests - Tool execution tests - Context and memory tests Closes #369 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com> * fix: capture fail tool arguments and add system prompt TODO comment - Capture fail tool arguments in error message instead of generic "Explicit fail called" - Line 233: Replace generic message with `inspect(lisp_step.return)` to preserve failure reason - Fixes PR review issue #2: failure data passed to fail tool is now preserved - Add comment linking to issue #374 for future system prompt enhancements - Line 313-314: Document that system prompt is intentionally minimal - Reference existing issue #374 for context-aware prompts, tool documentation, etc. - Fixes PR review issue #4: clarifies intentional design choice 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com> --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
5 tasks
This was referenced Dec 31, 2025
4 tasks
6 tasks
andreasronge
added a commit
that referenced
this pull request
Jan 11, 2026
- Add message-history-optimization-roadmap.md with 24 GitHub issues - Add migration-guard.md with compile-time enforcement hints - Add Implementation Phases section with incremental strategy - Add Cleanup Requirements (CLN-001 to CLN-012) - Add ARC-011, ARC-012: clarify compression scope vs prompt structure - Address critical risks: - "Blind LLM" gap: #18 prepares only, #22 does atomic switch - Breaking demo: #21 dual-write only, #24 does final cleanup - PTC-Lisp regressions: #2 requires edge case tests Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This was referenced Jan 24, 2026
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
This PR tests the newly added GitHub Actions workflows.
🤖 Generated with Claude Code