Open-source contributions
Every pull request opened from the lab account into someone else’s repository — open, closed and merged alike.
236pull requests
158repositories
143organisations
55open
99merged
82closed
99 merged.
This page is generated straight from the GitHub API by
tools/build_site.py,
so it cannot quietly drift from what is actually on GitHub, and nothing can be left off it to make the
record look better.
Code, recipes and evals into vendor & framework repositories
feat: Add sqlite-graph-memory to Developer Tools
fix(codex): use a ChatGPT-accepted default summarize model and mark failed summaries
fix(integrations): cancel the Hermes MCP actor once so a repeat shutdown cannot orphan bm mcp
fix(search): memory_search no longer returns nothing for question-shaped queries
Fix Hermes hook install steps that left the hook silently off
test: pin the auto hook's KeyboardInterrupt output end to end
fix(agent-setup): stop Grok hook errors outside Paradev terminals
health: flag a source that calls itself unverified but has no source_quality (#258)
fix(update-check): only notify when local is actually behind origin/main
fix(flow): free only the exact port on Windows, not every port it prefixes
Ignore CRLF line endings in the battery freeze check
save_note: return an error when a target parent is a file
move_note: return an error when a destination parent is a file
fix(fs): write, edit, apply_patch and mv replace a file atomically
fix silent play after pause+stop and after clicking a word while paused
Skip PyPy 3.9 wheels so the 17.0.0 build can publish
Tanner death blocks a Werewolf win
Update root SHA256SUMS.txt to the feedback-002 build
porkbun_dns: stop --mail from recreating TXT records on every run
Bash: fix all ShellCheck error-severity findings (part of #44)
Fix import without OPENAI_API_KEY and parse_json on escaped quotes
fix(close-call): build puzzles for six-digit NORAD objects; skip docked pairs
verify.py: free the first model before loading the second; read JSON as UTF-8
Fix MSVC build with std::pair / std::map recursive references (#3)
Sync package-lock.json with package.json so npm ci works
Add README with local setup steps
Fix _iv prompt id not detected at the end of SWE-bench task names
fix(httpx): keep the no-drain reuse control valid on go 1.27
fix(duet): drop leftover resume and session-economy probe guidance (v2.1.2)
Extract the Reid 2019 source with a relative tar path
Fix five broken links in four essays
Tests: the consumer-death stream test waits for a pool place, not 100 ms
Pin LF line endings so a Windows checkout keeps the pinned bytes
Make the fresh-clone quick start and test command work
Explain why sleep-switch toggle refuses an unknown status
check: spend the instance's daily budget only on admitted checks
Fix meta-level data file repeating rows for single-sample models (#124)
A refractor trigger keeps its target grouped (^^probe(1 + 2) * 3)
Cover ADL on chat datasets and fix the chat path it exposed
fix(tool-presentation): read a presentation or defaults file saved with a UTF-8 BOM (#758)
fix(ui): strip carriage returns from git status porcelain paths
docs: fix generate_report example arguments
Add sqlite-graph-memory MCP server
test(core): characterize leader delivery and terminal state per host class
fix(xml-tools): mark a parameter nullable only when its enum admits null (#150)
Fix move_note swapping note contents under cloud sync (#324)
fix(stdio): exit on SIGINT, SIGTERM and closed stdin regardless of session state (#17)
Fix the Jev Wikipedia link and add verified URLs to resources/repos.md
docs: point env reference links to the published page (docs/ENVS.md was moved)
Import the component files explicitly so case-insensitive filesystems resolve them
fix(webui): a subagent's alert rows clear once it leaves the health snapshot
Fix missing layer norms after re-running ADL on a new layer, and raise a clear error
fix(py): reject async trigger callbacks at registration
fix(gate): load gate.json and privileged-users files saved with a UTF-8 BOM (#395)
docs: fix three broken tutorial snippets (LM Studio URL, SQLChatAgent, unterminated string)
docs: fix stale caura_manage op list and moved document_repository path
Fix corpus cpWER aggregation across sessions
Expire an LLM job when its caller stops waiting
fix(scripts): run workspace checks on Windows
fix(inkbox): persist durable inbox on Windows
fix(policy): match escaped blocked merchant names
fix(compact): time-stamp the Memory Compact heading (#768)
deploy: copy the context manifest into the gateway image so it builds
fix: let compact write into a store directory not named memory
docs(security): sync runtime-env knob table with RUNTIME_ENV_KNOBS
docs(dev): fix llm_analyzer_base source links
fix(policy): blocked merchants are missed unless the name is a top-level key
ci: build before typecheck so the check job can pass
test(desktop): pin both Linux release legs to the ubuntu-22.04 glibc floor
proposals/ast-fixture-corpus: AST09 actor-tamper vector (requested in #81) + two JCS canonicalization vectors
docs: task-termination guide passes single_round/done_if_response to TaskConfig, where they are silently ignored
docs: fix broken link to the v0.4.0 Sift benchmark in README
fix(bedrock): keep every `nova_code_interpreter_result` toolResult content block
fix(mcp): name the capped server timeout in App read timeout warnings
fix(schedule): report a systemd timer with no next elapse as broken
fix(mcp): report the real checksum from write_note and edit_note (#1586)
docs: correct CLI availability of extension tools
docs: mark HeadroomClient provider as required, not auto-detected
fix(core): add opt-in failMode: "closed" for PreToolUse command hooks
fix(docs): harness docs document an async client API that does not exist
fix: merge generation_kwargs tools with the component's own tools in OpenAIChatGenerator
docs: fix dead github-cli link in CONTRIBUTING
docs(actor): fix drag_to parameter name in Element API reference
fix(acp): report max_tokens instead of end_turn on unresolved output truncation
fix(dashboard): install node dependencies before building
fix(mcp): resolve the API port from runtime.json instead of assuming 8000
feat: make per-tool extension allowlists configurable via TELEGRAM_FILE_EXTENSIONS (closes #226)
docs(llms): Sarvam sampling params go inside the model dict
docs(clients): fix CC system prompt override command (prompts print-cc-system-prompt-override)
docs: MCP tool count is 17; document get_file_outline, compare_graphs, check_index_coverage
fix(mcp): keep the bridge importable under the system python3 (3.9)
engine: sanitize bracketed text in Session::paste_text (#275)
feat(integrations): add support for huggingface
feat(integrations): add support for cohere
feat(integrations): add support for fireworks-ai
feat(integrations): add support for together-ai
feat(integrations): add support for groq
feat(integrations): add support for deepseek
feat(integrations): add support for mistral
fix(ai-proxy): keep empty anthropic tools array
feat(integrations): add support for lambda-cloud
feat(integrations): add support for openrouter
feat(core): accept SQLite URLs in database_url (#539)
fix(platform): discover KiCad's socket from metadata, never by connecting
fix(mcp): treat Talk's 304 Not Modified as an empty result
feat(model/openaimodel): support image/file input via genai InlineData/FileData parts
docs: fix six relative links that resolve to files that do not exist
docs: repoint doc links left behind by the bilingual docs split
feat(plugins): make the daily memory filename collision-safe on synced directories
[Github] Make code-format-helper --extensions ordering deterministic
feat(export): add RIS citation exporter for Zotero and Mendeley (#10)
fix(platform): record each server and sweep dead records at startup
fix(drc): name what owns each DRC report item (#413)
docs: fix Azure deploy paths after the v3 multi-package move
docs: point plugin setup links at the sections that replaced the removed guides
Python: pin the validated address for OpenAPI plugin requests
test(docker): exposed mode authenticates, and MARM's managed Docker path never reaches the keyless fallback (#170 item 5)
docs: trusty-memory and trusty-search are cargo crates, not uv tools
fix(placement): stop scoring connector filter caps as decoupling defects
test(docker): the healthcheck could never report unhealthy (#170 item 5)
Python: document connector extras and optional dependency compatibility
ci: build the image on PRs and run the docker-marked tests (#170 item 1)
docs: bake example used an npm package that does not exist
docs: point Claude Code MCP setup at the file Claude Code reads
feat(connector): Reddit data-source connector (cognee#4811)
feat(connector): Telegram data-source connector (cognee#4730)
feat(connector): Obsidian vault data-source connector (cognee#4725)
feat(mcp): surface ranker, scores, and invalidated-fact counts in search results (#1645)
fix(scripts): repair smoke_embedding_chunking against the v2.14.0 chunk profiles
docs(oauth): the "Cannot find module" fix suggested a broken import
docs(pypi): make the PyPI README's links absolute so they resolve
fix(cli): apply the documented `mex timeline --type` filter
docs: make the protocol.md table of contents resolve
docs: npx route for the mcp-remote bridge, with the config that matches it
MCP03: add plan-vs-authorize containment control + attack/defense reference
perf(hooks): refresh the PyPI check off the session-start blocking path (#676)
docs: mcp-remote bridge install names a package that does not exist on npm
fix(mcp): route listen-channel messages through the shared ping tally
fix(core): do not claim a question host in stream-json direct mode
docs: add discontinuation note to the Meta Llama API integration tile
fix(plugin): keep the server PID record correct across sessions and reap on exit
docs(install): install the CLI from the repo — book-to-skill is not on PyPI
fix(eval): record Task(approval=...) policies in the eval log config
docs: fix four dead links to the Multi Round-Trip Requests section
fix(edge): answer JSON-mode POSTs whose requests are never answered
docs: add a keyboard shortcuts reference (#36)
fix(apply-diff): accept the stacked @@ headers the tool description asks for
fix: repair four imports that do not resolve in cookbooks
ci: sample the stack when the skipped tests hang on a runner (#1)
Add the missing `cerebras` optional group to `pydantic-ai-slim`
fix(configurable): apply config_path containment to workflow node refs
Python: bind tool-approval responses to surfaced approval requests (#7383)
fix(scorer): decide a grader panel by strict majority, not mode
docs: drop install extras logfire does not have
docs: quick start `go get` leaves the module unbuildable
fix(trimmer): preview structured tool outputs from their text parts
docs: document the MCP_TRANSPORT=sse shared server (fixes the 'one instance per agent' surprise in #50)
docs: fix the advanced OAuth example so it compiles
docs: fix the streamable client snippet to use (*Client).Connect
fix(integrations): stop bm subprocesses inheriting Hermes's Python env
Add evals-tools plugin with judge-panel skill (cheap detector -> LLM judge triage)
[Content] Defending Grok Agents Against Prompt Injection in Tool Results
[Content] Evaluating Grok Applications: Deterministic Checks + LLM-as-Judge
[Content] Grounded RAG with Verifiable Citations (Grok)
read-arxiv-paper skill: handle the URL and archive shapes arxiv actually serves
Python: Add experimental FunctionAuthorizationFilter for auto function invocation (runtime authorization, argument-bound approvals)
fix: convert remaining broken sre_bot_slack.py links in SRE cookbook
Add plan-file-hygiene skill (addresses #1417)
Add example: check citation faithfulness in RAG
Add cookbook: check citation faithfulness in RAG with a zero-token gate
Add citation faithfulness check notebook (RAG)
Add Citation Faithfulness Check notebook (RAG cookbook)
feat(examples): deterministic guard against fabricated tool results (#737)
feat: add CitationConsistencyChecker — deterministic, zero-token RAG citation validator
Add self-verifying search agent recipe
feat: add AuthorityRoutingPlugin for ADVISE/EXECUTE/DEFER/STOP tool authority
feat(plugins): add AuthorityRoutingPlugin (ADVISE/EXECUTE/DEFER/STOP)
[Content] Deterministic Guardrails for Grok Tool Calls
feat(examples): RAG with verifiable citations for the Assistant
Add reasoning-quality-gate skill
feat(evals): citation-faithfulness evals — catching hallucinated support
Add verified-ops-agent: Python demo of propose → adversarially verify → gated apply
Add pipeline vs barrier sub-agent composition cookbook (addresses #721)
feat(agents): add authority routing pattern (ADVISE / EXECUTE / DEFER / STOP)
feat(agents): add multi-agent consensus & verification pattern
feat(patterns/agents): coordinating agents that don't share memory (message-bus consensus + liveness)
Entries into ecosystem lists
Watchlist: add agent-leash
Add sqlite-graph-memory to Model Context Protocol (MCP)
新增 sqlite-graph-memory 到 🧠 知识、记忆与 RAG
Add sqlite-graph-memory to Memory - Knowledge Management
Add sqlite-graph-memory to Agent Memory & State
Add Second Brain Starter Kit to Skill Collections
Add SQLite Graph Memory to Knowledge Management
Add tonydzi/sqlite-graph-memory to Knowledge & Memory 🤖🤖🤖
Add second-brain-starter-kit to Development & Workflow
Add tonydzi/sqlite-graph-memory to other
Add telegram-mcp-kit to Open Source Tools
Add second-brain-starter-kit to Tools
Add telegram-mcp-kit to Tools
Add tonydzi/second-brain-starter-kit to Skills
Add Second Brain Starter Kit to Claude Code plugins catalog
Add tonydzi/sqlite-graph-memory to mcp_server_repos.json
Add tonydzi/sqlite-graph-memory to Knowledge Management & Memory
Add sqlite-graph-memory (AI, Agents, and Memory)
Add sqlite-graph-memory to Development & Workflow
Add sqlite-graph-memory to MCP memory servers
Add sqlite-graph-memory to Local-First & Coding-Agent Memory
Add The Bible Framework (persona as versioned rules) to Memory & Identity
Product & Project: add sqlite-graph-memory (wikilink graph + SQLite, no graph DB)
watchlist: add CharmOS (humans + agents in one relationship record on a local vault)
Memory, State, and Communication: add claw-consensus (cross-machine consensus + dual-rail bus)
Add agent-leash (plan-vs-authorize containment) to Tools
Add claude-memory-tidy to Emerging projects (Open-Source)
Add mcp-daemon-diet to Tools
Add skill: tonydzi/second-brain-skills
Add claude-memory-tidy to Memory & Context Management
Guardrail Frameworks: add verbatim-citation-gate (deterministic, fail-closed)
Evaluation: add verbatim-citation-gate (deterministic citation check)
RAG Evaluation: add verbatim-citation-gate (deterministic pre-filter)
Benchmarks: add agent-runtime-integrity-bench (silent state-integrity violations in agent runtimes)
Configuration & Templates: add claude-bible
Agentic security: add LEASH-8 control model
Add sqlite-graph-memory to Emerging projects
Add SQLite Graph Memory to Memory tools
Add Verbatim Citation Gate to RAG
5c: add Verbatim Citation Gate (RAG / retrieval evaluation)
Add SQLite Graph Memory to Library
Add LEASH-8 to Industry Reports & Guidance
Blue-teaming: add LEASH-8 under Agent privilege management
Add SQLite Graph Memory to Advanced Approaches
Add Verbatim Citation Gate to Evaluation & Benchmarking
Add SQLite Graph Memory to Systems and Open Sources
Add claude-consensus (multi-machine agent consensus)
Add: dual-rail-message-delivery