1841 Commits

Author	SHA1	Message	Date
Claude (Opus 4.6)	a2a40429bd	Merge pull request '[claude] Poka-yoke: auto-revert incomplete skill edits (#923)' (#946) from claude/issue-923 into main	2026-04-21 16:38:24 +00:00
Alexander Whitestone	1fece10569	feat: poka-yoke auto-revert for incomplete skill edits (#923) ... Implement a transactional write-validate-commit-or-rollback pattern for all skill_manage write operations (edit, patch, write_file): - _backup_skill_file: timestamped .bak.{ts} snapshot before every write - _validate_written_file: re-reads from disk after write to catch truncation, encoding errors, and broken YAML frontmatter - _revert_from_backup: restores original content (or removes the corrupted file) on any validation failure - _cleanup_old_backups: prunes to MAX_BACKUPS_PER_FILE (3) after success; failed edits keep their .bak file as a debugging aid Also fixes pre-existing issue where _patch_skill error returns lacked a `suggestion` field expected by test_skill_manager_error_context.py tests. Adds 21 tests in test_skill_manager_autorevert.py covering every component and an end-to-end simulation of mid-write failure + auto-revert. Fixes #923 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-21 11:37:55 -04:00
Alexander Whitestone	46668505bc	Merge pull request 'feat: tool fixation detection — break repetitive loops (#886)' (#914) from fix/886 into main	2026-04-21 15:35:08 +00:00
Alexander Whitestone	cac0c8224e	Merge pull request 'fix: circuit breaker for error cascading (2.33x amplification)' (#927) from fix/885-circuit-breaker into main	2026-04-21 15:35:04 +00:00
Alexander Whitestone	f38a64455d	Merge pull request '[claude] Gateway config debt: add validation tests and API_SERVER_KEY warning (#892)' (#915) from claude/issue-892 into main	2026-04-21 15:33:19 +00:00
Alexander Whitestone	cf090a966d	Merge pull request 'fix: Poka-yoke — detect and block tool hallucination before API calls (#922)' (#935) from fix/922 into main	2026-04-21 15:29:35 +00:00
Alexander Whitestone	690d100afc	Merge pull request 'feat: Poka-yoke token budget — progressive context overflow guard (#925)' (#943) from burn/925-1776770102 into main	2026-04-21 15:29:02 +00:00
Alexander Whitestone	c6f0831738	Merge pull request 'feat: Python syntax validation before execute_code (#913)' (#917) from fix/913-syntax-validation into main	2026-04-21 15:27:05 +00:00
Alexander Whitestone	feb24bd08c	Merge pull request 'feat: Block silent credential exposure in tool outputs (#839)' (#910) from fix/839-1776403070 into main	2026-04-21 15:26:47 +00:00
Alexander Whitestone	bc55f40505	Merge pull request 'feat: time-aware model routing for cron jobs (#889)' (#909) from fix/889 into main	2026-04-21 15:26:43 +00:00
Alexander Whitestone	2adc72335e	Merge pull request 'fix: profile session isolation — tag and filter by profile' (#907) from fix/891-profile-isolation into main	2026-04-21 15:26:39 +00:00
Alexander Whitestone	ab32670464	Merge pull request 'feat: Poka-yoke — detect and block tool hallucination before API calls (#922)' (#944) from burn/922-1776770102 into main	2026-04-21 15:23:56 +00:00
Alexander Whitestone	27d2f2ca0e	Merge pull request 'feat: Prevent context window overflow via proactive token counting (#838)' (#905) from fix/838-1776402240 into main	2026-04-21 15:22:31 +00:00
Alexander Whitestone	08432a5618	test: poka-yoke validation tests (#922)	2026-04-21 11:59:26 +00:00
Alexander Whitestone	07c5b5b83d	test: add token budget poka-yoke tests (#925)	2026-04-21 11:41:39 +00:00
Alexander Whitestone	6eeee39c10	test(#922): Add tests for tool hallucination detection ... Tests for validation firewall: - Unknown tool detection - Missing required params - Wrong type detection - Hallucination patterns - Rejection stats Refs #922	2026-04-21 05:38:54 +00:00
Alexander Whitestone	30509b9c7c	test: circuit breaker tests ... Part of #885	2026-04-21 00:28:15 +00:00
Alexander Whitestone	c17f64fa2c	test: add syntax validation tests (#913)	2026-04-20 15:47:35 +00:00
Alexander Whitestone	c22cdcaa8e	fix: add _validate_gateway_config tests and API_SERVER_KEY network binding warning ... Refs #892 - Gateway config debt: missing keys and broken fallbacks Changes: - Add `_is_network_accessible()` helper to gateway/config.py (avoids circular import with gateway.platforms.base which imports from gateway.config) - Add API_SERVER_KEY warning in `_validate_gateway_config`: when the API server is enabled on a network-accessible address (0.0.0.0, public IP, hostname) but no key is configured, log a warning at config-load time so operators see the issue before any adapter initialisation runs - Add `TestValidateGatewayConfig` in tests/gateway/test_config.py covering: - idle_minutes <= 0 and None are corrected to 1440 (default) - at_hour outside 0-23 is corrected to 4 (default) - Boundary hours 0 and 23 are accepted unchanged - Empty platform token triggers a warning log - Disabled platform with empty token produces no warning - API server on 0.0.0.0 without key logs a warning - API server on 127.0.0.1 without key is silent (loopback is allowed) - API server with a key set logs no warning regardless of bind address Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-17 02:18:02 -04:00
Alexander Whitestone	ab968e910c	feat: tool fixation detection — break repetitive loops (#886) ... Marathon sessions show tool fixation: agent latches onto one tool and calls it repeatedly. Observed streaks of 8-25 identical calls. New agent/tool_fixation_detector.py: - ToolFixationDetector: tracks consecutive tool calls - record(tool_name): returns nudge prompt when threshold reached - Default threshold: 5 consecutive calls (configurable via TOOL_FIXATION_THRESHOLD env var) - Nudge prompt explains the fixation and suggests alternatives: 1. Read error carefully 2. Try different tool 3. Ask user for clarification 4. Check if task is complete - get_streak_info(): current streak state - format_report(): human-readable fixation events - Singleton via get_fixation_detector() Config: - TOOL_FIXATION_THRESHOLD (default: 5) - TOOL_FIXATION_WINDOW (default: 10) Tests: tests/test_tool_fixation_detector.py (9 tests) Closes #886	2026-04-17 01:57:37 -04:00
Alexander Whitestone	cb331da4f1	test: Add credential redaction tests (#839)	2026-04-17 05:23:48 +00:00
Alexander Whitestone	0b72884750	feat: time-aware model routing for cron jobs (#889) ... Error rate peaks at 18:00 (9.4%) during evening cron batches vs 4.0% at 09:00 during interactive work. Route cron tasks to stronger models during off-hours when user is not present to correct errors. New agent/time_aware_routing.py: - resolve_time_aware_model(): routes based on hour, error rate, task type - Interactive sessions: always use base model (user corrects errors) - Cron during business hours: use base model (low error rate) - Cron during off-hours with high error rate (>6%): upgrade to strong model - get_hour_error_rate(): error rates by hour from empirical audit - is_off_hours(): 18:00-05:59 = off-hours - RoutingDecision: model, provider, reason, hour, error_rate - get_routing_report(): 24h forecast of routing decisions Config via env vars: - CRON_STRONG_MODEL (default: xiaomi/mimo-v2-pro) - CRON_CHEAP_MODEL (default: qwen2.5:7b) - CRON_ERROR_THRESHOLD (default: 6.0%) Tests: tests/test_time_aware_routing.py (9 tests) Closes #889	2026-04-17 01:15:09 -04:00
Alexander Whitestone	a0ed1e6ff2	test: profile isolation tests ... Part of #891	2026-04-17 05:13:03 +00:00
Alexander Whitestone	d4cdfdc604	test: Add context budget tracker tests (#838)	2026-04-17 05:06:54 +00:00
Timmy Time	34e7de6a4c	feat: 988 Lifeline tests (#673)	2026-04-17 05:04:50 +00:00
Alexander Whitestone	05f8c2d188	Merge PR #899 ... Merged PR #899: feat: Allegro worker deliverables	2026-04-17 01:52:11 +00:00
Hermes Agent	ff2ce95ade	feat(research): Allegro worker deliverables — fleet research reports + skill manager test ... Research reports: - Vector DB research - Workflow orchestration research - Fleet knowledge graph SOTA research - LLM inference optimization - Local model crisis quality - Memory systems SOTA - Multi-agent coordination - R5 vs E2E gap analysis - Text-to-music-video Test: - test_skill_manager_error_context.py [Allegro] Forge workers — 2026-04-16	2026-04-16 15:04:28 +00:00
Hermes Merge Bot	aedebfdf58	Merge PR #848	2026-04-16 02:12:13 -04:00
Hermes Merge Bot	adf49b1809	Merge PR #849	2026-04-16 02:11:21 -04:00
Hermes Merge Bot	52ea3a8935	Merge PR #850	2026-04-16 02:09:00 -04:00
Hermes Merge Bot	43246d6cb4	Merge PR #852	2026-04-16 02:08:06 -04:00
Hermes Merge Bot	dff451081d	Merge PR #856	2026-04-16 02:05:42 -04:00
Hermes Merge Bot	5509b157c5	Merge PR #864	2026-04-16 02:05:05 -04:00
Hermes Merge Bot	fcc322fb81	Merge PR #867	2026-04-16 02:03:23 -04:00
Hermes Merge Bot	9bba9ecc40	Merge PR #866	2026-04-16 02:02:43 -04:00
Alexander Whitestone	3238cf4eb1	feat: Tool investigation report + Mem0 local provider (#842) ... ## Investigation Report - docs/tool-investigation-2026-04-15.md: Full report analyzing 414 tools from awesome-ai-tools. Top 5 recommendations with integration paths. - docs/plans/awesome-ai-tools-integration.md: Implementation tracking plan. ## Mem0 Local Provider (P1) - plugins/memory/mem0_local/: New ChromaDB-backed memory provider. No API key required - fully sovereign. Compatible tool schemas with cloud Mem0 (mem0_profile, mem0_search, mem0_conclude). - Pattern-based fact extraction from conversations. - Deterministic dedup via content hashing. - Circuit breaker for resilience. - tests/plugins/memory/test_mem0_local.py: Full test coverage. ## Issues Filed - #857: LightRAG integration (P2) - #858: n8n workflow orchestration (P3) - #859: RAGFlow document understanding (P4) - #860: tensorzero LLMOps evaluation (P3) Closes #842	2026-04-15 23:04:41 -04:00
Timmy	eed87e454e	test: Benchmark Gemma 4 vision accuracy vs current approach (#817) ... Vision benchmark suite comparing Gemma 4 (google/gemma-4-27b-it) vs current Gemini 3 Flash Preview (google/gemini-3-flash-preview). Metrics: - OCR accuracy (character + word overlap) - Description completeness (keyword coverage) - Structural quality (length, sentences, numbers) - Latency (ms per image) - Token usage - Consistency across runs Features: - 24 diverse test images (screenshots, diagrams, photos, charts) - Category-specific evaluation prompts - Automated verdict with composite scoring - JSON + markdown report output - 28 unit tests passing Usage: python benchmarks/vision_benchmark.py --images benchmarks/test_images.json python benchmarks/vision_benchmark.py --url https://example.com/img.png python benchmarks/vision_benchmark.py --generate-dataset Closes #817.	2026-04-15 23:02:02 -04:00
Alexander Whitestone	f03709aa29	test: crisis hook integration tests with agent loop (#707) ... 10 integration tests verifying crisis detection works correctly when called from the agent conversation flow: - scan_user_message detects CRITICAL/HIGH/MEDIUM/LOW levels - Safe messages pass through without triggering - Tool handler returns valid JSON - Compassion injection includes 988 lifeline for CRITICAL/HIGH - Case insensitive detection - Empty/None text handled gracefully - False positive resistance on common non-crisis phrases - Config check returns bool - Callable from agent context (not just isolation tests)	2026-04-15 23:00:12 -04:00
PRIMA	85a654348a	feat: poka-yoke — prevent hardcoded ~/.hermes paths (closes #835) ... scripts/lint_hardcoded_paths.py (new): - Scans Python files for hardcoded home-directory paths - Detects: Path.home()/.hermes without env fallback, /Users/<name>/, /home/<name>/ - Excludes: comments, docstrings, test files, skills, plugins, docs - Excludes correct patterns: profiles_parent, current_default, native_home - Supports --staged (git pre-commit), --fix (suggestions), --json output scripts/pre-commit-hardcoded-paths.sh (new): - Pre-commit hook that runs lint_hardcoded_paths.py --staged - Blocks commits containing hardcoded path violations tools/confirmation_daemon.py (fixed): - Replaced Path.home() / '.hermes' / 'approval_whitelist.json' with get_hermes_home() / 'approval_whitelist.json' - Added import of get_hermes_home from hermes_constants tests/test_hardcoded_paths.py (new): - 11 tests: detection, exclusion, fallback patterns, clean files	2026-04-15 22:56:32 -04:00
Alexander Whitestone	13ef670c05	feat: session compaction with fact extraction (#748) ... Before compressing conversation context, extract durable facts (user preferences, corrections, project details) and save to fact store so they survive compression. New agent/session_compactor.py: - extract_facts_from_messages(): scans user messages for preferences, corrections, project/infra facts using regex - 3 pattern categories: user_pref (5 patterns), correction (3 patterns), project (4 patterns) - ExtractedFact: category, entity, content, confidence, source_turn - save_facts_to_store(): saves to fact store (callback or auto-detect) - extract_and_save_facts(): one-call extraction + persistence - Deduplication by category+content - Skips tool results, short messages, system messages - format_facts_summary(): human-readable summary Tests: tests/test_session_compactor.py (9 tests) Closes #748	2026-04-15 22:41:54 -04:00
Alexander Whitestone	9f0c410481	feat: batch tool execution with parallel safety checks (#749) ... Centralized safety classification for tool call batches: tools/batch_executor.py (new): - classify_tool_calls() — classifies batch into parallel_safe, path_scoped, sequential, never_parallel tiers - BatchExecutionPlan — structured plan with parallel and sequential batches - Path conflict detection — write_file + patch on same file go sequential - Destructive command detection — rm, mv, sed -i, redirects - execute_parallel_batch() — ThreadPoolExecutor for concurrent execution tools/registry.py (enhanced): - ToolEntry.parallel_safe field — tools can declare parallel safety - registry.register() accepts parallel_safe=True parameter - registry.get_parallel_safe_tools() — query registry-declared safe tools Safety tiers: - parallel_safe: read_file, web_search, search_files, etc. - path_scoped: write_file, patch (concurrent when paths don't overlap) - sequential: terminal, delegate_task, unknown tools - never_parallel: clarify (requires user interaction) 19 tests passing.	2026-04-15 22:17:16 -04:00
Alexander Whitestone	b34b5b293d	test: add tests for tool hallucination prevention (#836)	2026-04-16 02:15:59 +00:00
Timmy Time	fb7464995c	fix: Ultraplan Mode for daily autonomous planning (closes #840)	2026-04-15 22:14:16 -04:00
Alexander Whitestone	7c71b7e73a	test: parallel tool calling — 2+ tools per response (#798)	2026-04-16 02:13:00 +00:00
Alexander Whitestone	4a3068b3b5	test: add regression tests for issue #834 KeyError fix	2026-04-16 02:12:36 +00:00
Alexander Whitestone	d8d7846897	feat: add tests/tools/test_confirmation_daemon.py from PR #397	2026-04-16 01:35:24 +00:00
Alexander Whitestone	6840d05554	feat: add tests/agent/test_privacy_filter.py from PR #397	2026-04-16 01:35:21 +00:00
Alexander Whitestone	30afd529ac	feat: add crisis detection tool — the-door integration (#141) ... New tool: tools/crisis_tool.py - Wraps the-door's canonical crisis detection (detect.py) - Scans user messages for despair/suicidal ideation - Classifies into NONE/LOW/MEDIUM/HIGH/CRITICAL tiers - Provides recommended actions per tier - Gateway hook: scan_user_message() for pre-API-call detection - System prompt injection: compassion_injection based on crisis level - Optional escalation logging to crisis_escalations.jsonl - Optional bridge API POST for HIGH+ (configurable via CRISIS_BRIDGE_URL) - Configurable via crisis_detection: true/false in config.yaml - Follows the-door design principles: never computes life value, never suggests death, errs on side of higher risk Also: tests/test_crisis_tool.py (9 tests, all passing)	2026-04-15 21:00:06 -04:00
asheriif	33ae403890	fix(gateway): fix matrix lingering typing indicator	2026-04-15 04:16:16 -07:00
Teknium	47e6ea84bb	fix: file handle bug, warning text, and tests for Discord media send ... - Fix file handle closed before POST: nest session.post() inside the 'with open()' block so aiohttp can read the file during upload - Update warning text to include weixin (also supports media delivery) - Add 8 unit tests covering: text+media, media-only, missing files, upload failures, multiple files, and _send_to_platform routing	2026-04-15 04:16:06 -07:00