feat: implement basic lesson system with metadata and auto-inclusion by TimeToBuildBob · Pull Request #687 · gptme/gptme

TimeToBuildBob · 2025-10-12T17:11:31Z

Implements Phase 1 of the enhanced lesson system.

Overview

Adds a basic lesson system to gptme that allows structured lessons with metadata to be automatically included in context based on keywords.

Core Components

Parser: Parse lessons with YAML frontmatter (match.keywords)
Index: Discover and index lessons from configured directories
Matcher: Match lessons based on context (message content, keywords)
Auto-include: Automatically include relevant lessons in context
Commands: CLI commands (/lesson list, /lesson search, /lesson show)

Features

YAML frontmatter with extensible match.keywords structure
Backward compatible with lessons without frontmatter
Inferred metadata (title, description, category)
Simple keyword matching for Phase 1
Ready for future extensions (globs, semantic matching)

Example Lesson Format

---
match:
  keywords: [patch, file, editing]
---

# Lesson Title

Lesson content...

Testing

Tested with gptme-bob workspace:

Successfully indexed 21 lessons
Parser handles lessons with and without frontmatter
Search and listing commands work correctly

Closes #686

Important

Implements a lesson system with metadata and auto-inclusion based on keywords, including a parser, index, matcher, and CLI commands, with comprehensive tests.

Behavior:
- Implements lesson system with metadata and auto-inclusion based on keywords.
- Adds CLI commands for lesson management: /lesson list, /lesson search, /lesson show, /lesson refresh.
- Supports YAML frontmatter for lessons with match.keywords.
- Backward compatible with lessons without frontmatter.
Core Components:
- parser.py: Parses lessons with YAML frontmatter, extracting metadata.
- index.py: Discovers and indexes lessons from directories.
- matcher.py: Matches lessons based on context using keywords.
- auto_include.py: Automatically includes relevant lessons in context.
Testing:
- Tests for parsing, indexing, matching, and CLI commands in tests/ directory.
- Mocking and fixtures used for testing various scenarios.

^{This description was created by}^{for dca167b. You can customize this summary. It will automatically update as commits are pushed.}

Implements Phase 1 of the enhanced lesson system (#686): Core Components: - Parser: Parse lessons with YAML frontmatter (match.keywords) - Index: Discover and index lessons from configured directories - Matcher: Match lessons based on context (message content, keywords) - Auto-include: Automatically include relevant lessons in context - Commands: CLI commands (/lesson list, /lesson search, /lesson show) Features: - YAML frontmatter with extensible match.keywords structure - Backward compatible with lessons without frontmatter - Inferred metadata (title, description, category) - Simple keyword matching for Phase 1 - Ready for future extensions (globs, semantic matching) Example lesson format: --- match: keywords: [patch, file, editing] --- # Lesson Title Lesson content... Tested with gptme-bob workspace lessons (21 lessons indexed). Related: #686

ellipsis-dev

Important

Looks good to me! 👍

Reviewed everything up to 4843011 in 52 seconds. Click for details.

Reviewed 728 lines of code in 6 files
Skipped 0 files when reviewing.
Skipped posting 6 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. gptme/lessons/auto_include.py:62

Draft comment:
Consider explicitly annotating the type of 'matches' (e.g. list of MatchResult) to improve clarity.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

2. gptme/lessons/auto_include.py:83

Draft comment:
Consider narrowing the exception handling or logging the full traceback to improve debugging in case of errors.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

3. gptme/lessons/commands.py:140

Draft comment:
For lesson search, consider paginating results or providing additional navigation if more than 10 lessons match.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

4. gptme/lessons/index.py:119

Draft comment:
Consider using case-insensitive comparisons in get_by_category and find_by_keywords to ensure consistent matching behavior.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

5. gptme/lessons/matcher.py:56

Draft comment:
Document and/or add tests for future matching criteria (e.g., tools, glob, semantic) to guide further development.
Reason this comment was not posted:
Confidence changes required: 20% <= threshold 50% None

6. gptme/lessons/parser.py:106

Draft comment:
Validate that the 'keywords' extracted from YAML frontmatter is a list to prevent potential type issues if the data is not formatted as expected.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

Workflow ID: wflow_DuzD3o3Nz5ds5rGG

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

codecov · 2025-10-12T17:15:38Z

Codecov Report

❌ Patch coverage is 78.40617% with 84 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
gptme/lessons/auto_include.py	14.58%	41 Missing ⚠️
gptme/lessons/index.py	42.85%	36 Missing ⚠️
gptme/tools/lessons.py	95.58%	3 Missing ⚠️
gptme/lessons/commands.py	98.16%	2 Missing ⚠️
gptme/lessons/parser.py	96.42%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!

- Implements lesson system as a ToolSpec following gptme tool pattern - Uses MESSAGE_PRE_PROCESS hook for automatic lesson inclusion - Registers /lesson command through tool system - Imports from gptme.lessons module for core functionality - Auto-discovered by tool loading system - Configuration via environment variables: - GPTME_LESSONS_AUTO_INCLUDE (default: true) - GPTME_LESSONS_MAX_INCLUDED (default: 5) Completes the lesson system integration with gptme's tool architecture.

ellipsis-dev

Caution

Changes requested ❌

Reviewed dc653f6 in 1 minute and 18 seconds. Click for details.

Reviewed 178 lines of code in 1 files
Skipped 0 files when reviewing.
Skipped posting 1 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. gptme/tools/lessons.py:108

Draft comment:
Ensure that 'match.matched_by' is always a list of strings; consider adding type-checking or a default to prevent join errors.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

Workflow ID: wflow_3RQiWfVMy2x1sAZs

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

gptme/tools/lessons.py

- Add test_lessons_commands.py with 31 tests covering all command functionality - Add test_lessons_tool.py with 20 tests covering tool integration - Coverage improvements: - gptme/lessons/commands.py: 13% → 98% - gptme/tools/lessons.py: 24% → 96% - Overall lesson system: 24% → 61% Tests cover: - Command parsing and dispatch (list, search, show, refresh) - Auto-include hooks with configuration - Lesson formatting and display - Error handling and edge cases - Tool specification and registration

ellipsis-dev

Important

Looks good to me! 👍

Reviewed 8a28b6d in 50 seconds. Click for details.

Reviewed 753 lines of code in 2 files
Skipped 0 files when reviewing.
Skipped posting 6 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. tests/test_lessons_commands.py:22

Draft comment:
Ensure the type hint 'list[str] | None' is supported on your targeted Python version (3.10+); otherwise, consider a compatible syntax.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

2. tests/test_lessons_commands.py:208

Draft comment:
The test for search limits relies on exact formatting ('... and 5 more'). Consider using a regex or more flexible assertion to avoid brittleness if formatting changes.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

3. tests/test_lessons_commands.py:386

Draft comment:
Verify that error messages for unknown subcommands remain consistent between tests and implementation.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

4. tests/test_lessons_tool.py:149

Draft comment:
In test_hook_limits_max_lessons, the test relies on exact string matching for lesson titles. Consider abstracting the formatting to reduce brittleness in case the output format changes.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

5. tests/test_lessons_tool.py:197

Draft comment:
Consider adding a test case where lessons exist and an invalid max lessons config value is provided, to verify that the default (e.g., 5) is used instead of returning no output.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

6. tests/test_lessons_tool.py:327

Draft comment:
Using module reload in test_tool_availability_depends_on_module is effective but ensure it does not impact global state for other tests.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

Workflow ID: wflow_q5sJLVQ5tU9PgwyM

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

- Add test_lessons_parser.py with 28 tests covering parser functionality - Coverage improvement: gptme/lessons/parser.py: 36% → 96% - Overall lesson system: 61% → 69% (approaching 70% goal) Tests cover: - Title extraction from markdown - Description extraction with edge cases - Lesson parsing with/without YAML frontmatter - Error handling (file not found, invalid YAML) - Unicode content and complex structures - Dataclass validation

ellipsis-dev

Important

Looks good to me! 👍

Reviewed 70819a6 in 58 seconds. Click for details.

Reviewed 376 lines of code in 1 files
Skipped 0 files when reviewing.
Skipped posting 7 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. tests/test_lessons_parser.py:16

Draft comment:
TestExtractTitle covers various cases (whitespace, missing header) effectively.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

2. tests/test_lessons_parser.py:50

Draft comment:
TestExtractDescription robustly verifies paragraph extraction and skipping empty lines.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

3. tests/test_lessons_parser.py:123

Draft comment:
test_parse_lesson_with_frontmatter correctly parses and strips YAML frontmatter and extracts metadata.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

4. tests/test_lessons_parser.py:196

Draft comment:
test_parse_lesson_invalid_yaml robustly checks error handling for malformed YAML frontmatter.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

5. tests/test_lessons_parser.py:215

Draft comment:
Simulating the absence of PyYAML with patch is a clever approach to test dependency error handling.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

6. tests/test_lessons_parser.py:316

Draft comment:
The malformed frontmatter test confirms fallback to regular content parsing; consider additional tests for varied malformed delimiter scenarios.
Reason this comment was not posted:
Confidence changes required: 33% <= threshold 50% None

7. tests/test_lessons_parser.py:339

Draft comment:
Dataclass tests effectively ensure proper initialization of Lesson and LessonMetadata.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

Workflow ID: wflow_xA4m4LTXICk8umdB

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

- Add test_lessons_matcher.py with 13 tests covering matching functionality - Coverage improvement: gptme/lessons/matcher.py: 42% → 100% - Overall lesson system: 69% → 75% (exceeded 70% goal!) Tests cover: - Keyword-based matching with scoring - Case-insensitive matching - Threshold-based filtering - Score-based sorting - Multiple lesson matching - Explicit keyword matching Final PR #687 status: - Created 92 comprehensive tests (31+20+28+13) - Improved coverage from 24% to 75% - Five modules at 96-100% coverage

ellipsis-dev

Important

Looks good to me! 👍

Reviewed 3ec306b in 32 seconds. Click for details.

Reviewed 237 lines of code in 1 files
Skipped 0 files when reviewing.
Skipped posting 2 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. tests/test_lessons_matcher.py:136

Draft comment:
Consider asserting the exact expected score (e.g. 3.0) instead of using a '> 2.0' check for clarity in the threshold test.
Reason this comment was not posted:
Confidence changes required: 50% <= threshold 50% None

2. tests/test_lessons_matcher.py:1

Draft comment:
Overall, the tests provide comprehensive coverage. As an improvement suggestion, consider using parameterized tests to reduce duplication for similar matching scenarios.
Reason this comment was not posted:
Confidence changes required: 30% <= threshold 50% None

Workflow ID: wflow_K2af3np2vWbrcTlk

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

gptme/lessons/commands.py

gptme/tools/lessons.py

ErikBjare

Looks okay for merge, but should fix the review comments I left. I think some of the tests are excessive and unnecessary, you were really coverage-maxing to a fault I think.

- Extract and display code suggestions from review comment bodies - Rewrite tests to use real GitHub API calls instead of mocking - Test against actual PR #687 with review comments and suggestions - Remove mock dependencies from tests All tests pass with real API validation.

…ead_pr tool Comprehensive enhancement of GitHub PR reading capabilities: 1. Code Context Snippets (gh.py) - Extract diff_hunk from review comments - Display 3-10 lines of code context around commented lines - Remove diff markers for clean display 2. Code Suggestion Extraction (gh.py) - Detect and parse ```suggestion blocks from comment bodies - Format with syntax highlighting 3. New read_pr Tool Command (tools/gh.py) - Single command for agents to get full PR context - Syntax: gh read_pr <pr_url> - Returns all PR info: description, comments, review comments, suggestions - Proper error handling for invalid URLs 4. Comprehensive Testing - Real GitHub API tests (no mocking) - Tests against actual PR #687 - 7 tests covering all functionality - Proper type handling for ExecuteFunc Closes #688

- Use logger.exception instead of logger.warning for better debugging - Remove duplicated help text from lesson() docstring - Remove command usage from tool instructions (not needed by LLM)

ellipsis-dev

Important

Looks good to me! 👍

Reviewed 99ec82e in 44 seconds. Click for details.

Reviewed 52 lines of code in 2 files
Skipped 0 files when reviewing.
Skipped posting 3 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. gptme/lessons/commands.py:16

Draft comment:
The docstring has been simplified. Ensure that detailed usage examples (now in _lesson_help) are sufficient for users.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

2. gptme/lessons/commands.py:148

Draft comment:
Redundant manual command listing in the help text was removed. Confirm that _lesson_help now fully covers usage details.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

3. gptme/tools/lessons.py:90

Draft comment:
Switching from logger.warning to logger.exception improves error diagnostics by logging the stack trace.
Reason this comment was not posted:
Confidence changes required: 0% <= threshold 50% None

Workflow ID: wflow_GDcIEkVXjdCN7Cb6

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

openhands-ai · 2025-10-13T10:19:10Z

Looks like there are a few issues preventing this PR from being merged!

GitHub Actions are failing:
- Test

If you'd like me to help, just leave a comment, like

@OpenHands please fix the failing actions on PR #687 at branch `feat/lesson-system`

Feel free to include any additional details that might help me get this PR into a better state.

_{^{You can manage your notification settings}}

The test was checking for /lesson command in instructions, but that was removed per review feedback (commands are for users, not LLM). Updated test to check for relevant content instead.

ellipsis-dev

Important

Looks good to me! 👍

Reviewed dca167b in 53 seconds. Click for details.

Reviewed 14 lines of code in 1 files
Skipped 0 files when reviewing.
Skipped posting 2 draft comments. View those below.
Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. tests/test_lessons_tool.py:322

Draft comment:
Updated assertion to check for 'lesson system' in lower-case instructions; ensures instructions detail the lesson system. Verify this text remains consistent with future changes.
Reason this comment was not posted:
Comment did not seem useful. Confidence is useful = 0% <= threshold 50% The comment is asking the PR author to verify that the text remains consistent with future changes, which is against the rules. It is not making a specific code suggestion or asking for a test to be written. Therefore, this comment should be removed.

2. tests/test_lessons_tool.py:324

Draft comment:
Added assertion to verify 'keywords' appear in instructions; clearly indicates metadata support. This aligns with the enhanced lesson system requirements.
Reason this comment was not posted:
Comment did not seem useful. Confidence is useful = 0% <= threshold 50% This comment seems to be purely informative, explaining what was done and why, without suggesting any changes or asking for confirmation. It doesn't align with the rules for useful comments.

Workflow ID: wflow_a094OdzHKqKtcgrt

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

* feat: add code context snippets to GitHub PR review comments Enhances get_github_pr_content() to include code context from diff_hunk in review comments, making it easier to understand what code is being commented on. - Extract diff_hunk from review comment API responses - Format and display code context with syntax highlighting - Remove diff markers (@@, +, -) for cleaner display - Add comprehensive tests with mocked gh CLI responses Closes #688 * feat: add code suggestion extraction and real API tests - Extract and display code suggestions from review comment bodies - Rewrite tests to use real GitHub API calls instead of mocking - Test against actual PR #687 with review comments and suggestions - Remove mock dependencies from tests All tests pass with real API validation. * feat: enhance GitHub PR reading with code context, suggestions, and read_pr tool Comprehensive enhancement of GitHub PR reading capabilities: 1. Code Context Snippets (gh.py) - Extract diff_hunk from review comments - Display 3-10 lines of code context around commented lines - Remove diff markers for clean display 2. Code Suggestion Extraction (gh.py) - Detect and parse ```suggestion blocks from comment bodies - Format with syntax highlighting 3. New read_pr Tool Command (tools/gh.py) - Single command for agents to get full PR context - Syntax: gh read_pr <pr_url> - Returns all PR info: description, comments, review comments, suggestions - Proper error handling for invalid URLs 4. Comprehensive Testing - Real GitHub API tests (no mocking) - Tests against actual PR #687 - 7 tests covering all functionality - Proper type handling for ExecuteFunc Closes #688 * refactor: rename read_pr to "gh pr view" to match gh CLI convention Makes the command more intuitive by following the standard gh CLI naming: - Old: gh read_pr <url> - New: gh pr view <url> This matches the native gh CLI command while providing enhanced output with code context, suggestions, and comprehensive review information. * feat: add "gh pr checks" to wait for CI completion Adds a new command to wait for all GitHub Actions checks to complete: - gh pr checks <pr_url> Features: - Polls check status every 10 seconds - Shows status updates only when status changes (less spammy) - Displays counts with emoji indicators (✅ passed, ❌ failed, 🔄 in progress) - Exits when all checks complete - Shows final summary of results Much cleaner than "gh run watch" and monitors all checks at once. * test: add comprehensive mocked tests for GitHub PR code context and suggestions - Created 8 mocked tests covering all new functionality - Tests code context extraction from diff_hunk - Tests code suggestion extraction from comment bodies - Tests edge cases (empty diff_hunk, multiple suggestions, malformed blocks) - Tests file extension extraction for syntax highlighting - Combined with existing tests achieves 73% coverage (up from 0% patch coverage) Addresses Codecov feedback on PR #689 * fix: address review comments from ellipsis-dev - Handle files without extensions (e.g., Makefile) by defaulting to "text" lang - Map unmapped check conclusions to "success" instead of undefined "completed" Addresses review comments from ellipsis-dev bot. * fix: show line ranges for multi-line review comments Display "path:start-end" for multi-line comments instead of just "path:end" to accurately reflect the lines being commented on, matching GitHub UI. Example: gptme/tools/lessons.py:138-155 instead of :155 * fix: use correct JSON schema type for gh tool parameter Change type from 'str' to 'string' to match JSON schema specification. Fixes OpenAI API schema validation error. * fix: handle gh CLI authentication errors in tests Skip test_gh_tool_read_pr when gh CLI is not authenticated, which is common in CI environments. This prevents false failures when gh is installed but not configured with credentials. * fix: use original_line as fallback for review comment line numbers GitHub API returns null for 'line' and 'start_line' when comments are on old commits. Use 'original_line' and 'original_start_line' as fallbacks to show the line numbers where the comment was originally made. --------- Co-authored-by: Erik Bjäreholt <erik@bjareho.lt>

- Created docs/lessons.rst with complete documentation - Added lessons to docs index - Enhanced docs/lessons/README.md with: - Autonomous mode configuration - CLI commands reference - Best practices for writing lessons - Migration guide for existing lessons - Added lessons feature to main README Post-merge documentation for PRs #687 and #722. Co-authored-by: Bob <bob@superuserlabs.org>

* docs: add comprehensive lesson system documentation - Created docs/lessons.rst with complete documentation - Added lessons to docs index - Enhanced docs/lessons/README.md with: - Autonomous mode configuration - CLI commands reference - Best practices for writing lessons - Migration guide for existing lessons - Added lessons feature to main README Post-merge documentation for PRs #687 and #722. Co-authored-by: Bob <bob@superuserlabs.org> * docs: fix lesson system documentation to match implementation Remove references to non-existent autonomous mode configuration variables (GPTME_LESSONS_AUTO_INCLUDE_AUTONOMOUS, GPTME_LESSONS_MAX_INCLUDED_AUTONOMOUS) that were removed during refactoring. The system now uses unified configuration for all modes: - GPTME_LESSONS_AUTO_INCLUDE (default: true) - GPTME_LESSONS_MAX_INCLUDED (default: 5) Keywords are extracted from both user and assistant messages in all contexts, with no mode detection. Fixes inconsistencies identified in Greptile review (PR #771). Co-authored-by: Bob <bob@superuserlabs.org> --------- Co-authored-by: Erik Bjäreholt <erik.bjareholt@gmail.com>

…ptme#687) * feat: implement basic lesson system with metadata and auto-inclusion Implements Phase 1 of the enhanced lesson system (gptme#686): Core Components: - Parser: Parse lessons with YAML frontmatter (match.keywords) - Index: Discover and index lessons from configured directories - Matcher: Match lessons based on context (message content, keywords) - Auto-include: Automatically include relevant lessons in context - Commands: CLI commands (/lesson list, /lesson search, /lesson show) Features: - YAML frontmatter with extensible match.keywords structure - Backward compatible with lessons without frontmatter - Inferred metadata (title, description, category) - Simple keyword matching for Phase 1 - Ready for future extensions (globs, semantic matching) Example lesson format: --- match: keywords: [patch, file, editing] --- # Lesson Title Lesson content... Tested with gptme-bob workspace lessons (21 lessons indexed). Related: gptme#686 * feat: add lessons as a proper tool with hooks and commands - Implements lesson system as a ToolSpec following gptme tool pattern - Uses MESSAGE_PRE_PROCESS hook for automatic lesson inclusion - Registers /lesson command through tool system - Imports from gptme.lessons module for core functionality - Auto-discovered by tool loading system - Configuration via environment variables: - GPTME_LESSONS_AUTO_INCLUDE (default: true) - GPTME_LESSONS_MAX_INCLUDED (default: 5) Completes the lesson system integration with gptme's tool architecture. * test: add comprehensive tests for lesson commands and tool - Add test_lessons_commands.py with 31 tests covering all command functionality - Add test_lessons_tool.py with 20 tests covering tool integration - Coverage improvements: - gptme/lessons/commands.py: 13% → 98% - gptme/tools/lessons.py: 24% → 96% - Overall lesson system: 24% → 61% Tests cover: - Command parsing and dispatch (list, search, show, refresh) - Auto-include hooks with configuration - Lesson formatting and display - Error handling and edge cases - Tool specification and registration * test: add comprehensive parser tests - Add test_lessons_parser.py with 28 tests covering parser functionality - Coverage improvement: gptme/lessons/parser.py: 36% → 96% - Overall lesson system: 61% → 69% (approaching 70% goal) Tests cover: - Title extraction from markdown - Description extraction with edge cases - Lesson parsing with/without YAML frontmatter - Error handling (file not found, invalid YAML) - Unicode content and complex structures - Dataclass validation * test: add comprehensive matcher tests, achieve 75% coverage - Add test_lessons_matcher.py with 13 tests covering matching functionality - Coverage improvement: gptme/lessons/matcher.py: 42% → 100% - Overall lesson system: 69% → 75% (exceeded 70% goal!) Tests cover: - Keyword-based matching with scoring - Case-insensitive matching - Threshold-based filtering - Score-based sorting - Multiple lesson matching - Explicit keyword matching Final PR gptme#687 status: - Created 92 comprehensive tests (31+20+28+13) - Improved coverage from 24% to 75% - Five modules at 96-100% coverage * fix: address review comments - Use logger.exception instead of logger.warning for better debugging - Remove duplicated help text from lesson() docstring - Remove command usage from tool instructions (not needed by LLM) * fix: update test to match new tool instructions The test was checking for /lesson command in instructions, but that was removed per review feedback (commands are for users, not LLM). Updated test to check for relevant content instead. --------- Co-authored-by: Erik Bjäreholt <erik@bjareho.lt>

…tme#689) * feat: add code context snippets to GitHub PR review comments Enhances get_github_pr_content() to include code context from diff_hunk in review comments, making it easier to understand what code is being commented on. - Extract diff_hunk from review comment API responses - Format and display code context with syntax highlighting - Remove diff markers (@@, +, -) for cleaner display - Add comprehensive tests with mocked gh CLI responses Closes gptme#688 * feat: add code suggestion extraction and real API tests - Extract and display code suggestions from review comment bodies - Rewrite tests to use real GitHub API calls instead of mocking - Test against actual PR gptme#687 with review comments and suggestions - Remove mock dependencies from tests All tests pass with real API validation. * feat: enhance GitHub PR reading with code context, suggestions, and read_pr tool Comprehensive enhancement of GitHub PR reading capabilities: 1. Code Context Snippets (gh.py) - Extract diff_hunk from review comments - Display 3-10 lines of code context around commented lines - Remove diff markers for clean display 2. Code Suggestion Extraction (gh.py) - Detect and parse ```suggestion blocks from comment bodies - Format with syntax highlighting 3. New read_pr Tool Command (tools/gh.py) - Single command for agents to get full PR context - Syntax: gh read_pr <pr_url> - Returns all PR info: description, comments, review comments, suggestions - Proper error handling for invalid URLs 4. Comprehensive Testing - Real GitHub API tests (no mocking) - Tests against actual PR gptme#687 - 7 tests covering all functionality - Proper type handling for ExecuteFunc Closes gptme#688 * refactor: rename read_pr to "gh pr view" to match gh CLI convention Makes the command more intuitive by following the standard gh CLI naming: - Old: gh read_pr <url> - New: gh pr view <url> This matches the native gh CLI command while providing enhanced output with code context, suggestions, and comprehensive review information. * feat: add "gh pr checks" to wait for CI completion Adds a new command to wait for all GitHub Actions checks to complete: - gh pr checks <pr_url> Features: - Polls check status every 10 seconds - Shows status updates only when status changes (less spammy) - Displays counts with emoji indicators (✅ passed, ❌ failed, 🔄 in progress) - Exits when all checks complete - Shows final summary of results Much cleaner than "gh run watch" and monitors all checks at once. * test: add comprehensive mocked tests for GitHub PR code context and suggestions - Created 8 mocked tests covering all new functionality - Tests code context extraction from diff_hunk - Tests code suggestion extraction from comment bodies - Tests edge cases (empty diff_hunk, multiple suggestions, malformed blocks) - Tests file extension extraction for syntax highlighting - Combined with existing tests achieves 73% coverage (up from 0% patch coverage) Addresses Codecov feedback on PR gptme#689 * fix: address review comments from ellipsis-dev - Handle files without extensions (e.g., Makefile) by defaulting to "text" lang - Map unmapped check conclusions to "success" instead of undefined "completed" Addresses review comments from ellipsis-dev bot. * fix: show line ranges for multi-line review comments Display "path:start-end" for multi-line comments instead of just "path:end" to accurately reflect the lines being commented on, matching GitHub UI. Example: gptme/tools/lessons.py:138-155 instead of :155 * fix: use correct JSON schema type for gh tool parameter Change type from 'str' to 'string' to match JSON schema specification. Fixes OpenAI API schema validation error. * fix: handle gh CLI authentication errors in tests Skip test_gh_tool_read_pr when gh CLI is not authenticated, which is common in CI environments. This prevents false failures when gh is installed but not configured with credentials. * fix: use original_line as fallback for review comment line numbers GitHub API returns null for 'line' and 'start_line' when comments are on old commits. Use 'original_line' and 'original_start_line' as fallbacks to show the line numbers where the comment was originally made. --------- Co-authored-by: Erik Bjäreholt <erik@bjareho.lt>

* docs: add comprehensive lesson system documentation - Created docs/lessons.rst with complete documentation - Added lessons to docs index - Enhanced docs/lessons/README.md with: - Autonomous mode configuration - CLI commands reference - Best practices for writing lessons - Migration guide for existing lessons - Added lessons feature to main README Post-merge documentation for PRs gptme#687 and gptme#722. Co-authored-by: Bob <bob@superuserlabs.org> * docs: fix lesson system documentation to match implementation Remove references to non-existent autonomous mode configuration variables (GPTME_LESSONS_AUTO_INCLUDE_AUTONOMOUS, GPTME_LESSONS_MAX_INCLUDED_AUTONOMOUS) that were removed during refactoring. The system now uses unified configuration for all modes: - GPTME_LESSONS_AUTO_INCLUDE (default: true) - GPTME_LESSONS_MAX_INCLUDED (default: 5) Keywords are extracted from both user and assistant messages in all contexts, with no mode detection. Fixes inconsistencies identified in Greptile review (PR gptme#771). Co-authored-by: Bob <bob@superuserlabs.org> --------- Co-authored-by: Erik Bjäreholt <erik.bjareholt@gmail.com>

- logmanager: count JSONL lines instead of loading entire file text - prompts: fix "seperately" typo - test_util_gh: fix flaky assertions for PR #687 (all comments now resolved)

PR #687's review threads are now all resolved (GraphQL isResolved=true, despite REST API showing resolved_at=null). Updated test to: - Assert "Unresolved" section is NOT present (was asserting it IS) - Add new test using PR #1489 which has actual unresolved threads - Document the REST/GraphQL API inconsistency

…#1496) * fix: cleanup test assertions, typo, and memory-efficient msg counting - logmanager: count JSONL lines instead of loading entire file text - prompts: fix "seperately" typo - test_util_gh: fix flaky assertions for PR #687 (all comments now resolved) * fix(test): update gh util test for resolved PR #687 review threads PR #687's review threads are now all resolved (GraphQL isResolved=true, despite REST API showing resolved_at=null). Updated test to: - Assert "Unresolved" section is NOT present (was asserting it IS) - Add new test using PR #1489 which has actual unresolved threads - Document the REST/GraphQL API inconsistency * fix(test): use PR #271 for unresolved review thread test Replace soft if-based assertion with hard assert using PR #271 (4 unresolved threads from Nov 2024, stable target). PR #1489 had all threads resolved, making the previous test a no-op. --------- Co-authored-by: TimeToBuildBob <TimeToBuildBob@users.noreply.github.com>

ellipsis-dev bot reviewed Oct 12, 2025

View reviewed changes

gptme/tools/lessons.py Outdated Show resolved Hide resolved

ellipsis-dev bot reviewed Oct 13, 2025

View reviewed changes

TimeToBuildBob mentioned this pull request Oct 13, 2025

Enhance lesson system with structured metadata, triggers, and skills integration #686

Closed

51 tasks

ErikBjare reviewed Oct 13, 2025

View reviewed changes

gptme/lessons/commands.py Outdated Show resolved Hide resolved

ErikBjare reviewed Oct 13, 2025

View reviewed changes

gptme/tools/lessons.py Outdated Show resolved Hide resolved

ErikBjare reviewed Oct 13, 2025

View reviewed changes

TimeToBuildBob mentioned this pull request Oct 13, 2025

feat: enhance GitHub PR reading with code context and suggestions #689

Merged

fix: address review comments

99ec82e

- Use logger.exception instead of logger.warning for better debugging - Remove duplicated help text from lesson() docstring - Remove command usage from tool instructions (not needed by LLM)

ellipsis-dev bot reviewed Oct 13, 2025

View reviewed changes

fix: update test to match new tool instructions

dca167b

The test was checking for /lesson command in instructions, but that was removed per review feedback (commands are for users, not LLM). Updated test to check for relevant content instead.

ellipsis-dev bot reviewed Oct 13, 2025

View reviewed changes

ErikBjare merged commit 3e81a0d into master Oct 13, 2025
10 checks passed

This was referenced Oct 15, 2025

feat(lessons): add mode-specific configuration for lesson auto-inclusion #706

Closed

feat(lessons): add message pattern-based mode detection for Phase 3 #722

Merged

This was referenced Oct 22, 2025

docs: add comprehensive lesson system documentation #771

Merged

docs(lessons): add Phase 6 comprehensive documentation #795

Merged

TimeToBuildBob mentioned this pull request Nov 18, 2025

feat: add restart tool #853

Merged

TimeToBuildBob mentioned this pull request Feb 26, 2026

fix: cleanup test assertions, typo, and memory-efficient msg counting #1496

Merged

2 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: implement basic lesson system with metadata and auto-inclusion#687

feat: implement basic lesson system with metadata and auto-inclusion#687
ErikBjare merged 7 commits intomasterfrom
feat/lesson-system

TimeToBuildBob commented Oct 12, 2025 •

edited by ellipsis-dev bot

Loading

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

codecov bot commented Oct 12, 2025 •

edited

Loading

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

Uh oh!

Uh oh!

ErikBjare left a comment

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

openhands-ai bot commented Oct 13, 2025

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

TimeToBuildBob commented Oct 12, 2025 • edited by ellipsis-dev bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Overview

Core Components

Features

Example Lesson Format

Testing

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

codecov bot commented Oct 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

ErikBjare left a comment

Choose a reason for hiding this comment

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

openhands-ai bot commented Oct 13, 2025

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

TimeToBuildBob commented Oct 12, 2025 •

edited by ellipsis-dev bot

Loading

codecov bot commented Oct 12, 2025 •

edited

Loading