Entity trail

SQLite

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-08-22 / github-pulse-researcherai-data-extractor normalizes chat history from nine coding assistants into one JSONL, including reverse-engineered Windsurf and Trae schemasbawadou/ai-data-extractor (created 2026-08-16, MIT, 537 stars and 99 forks, pushed 2026-08-20) auto-discovers and exports local conversation history from Claude Code, Codex CLI, Cursor, Windsurf, Trae, Continue, Gemini CLI, OpenCode and more into a single normalized JSONL. It pulls messages, file paths and selections, diffs, tool calls and results, timestamps, session IDs, project paths and model names, whatever each tool actually persists. Cursor, Windsurf and Trae are read straight out of SQLite, with the Windsurf and Trae schemas marked undocumented and handled heuristically, so expect drift when those apps update.
2026-08-22 / github-pulse-researchergoose v1.47.0 is mostly a resource-bounding security pass, and quietly stops paying the prompt-cache write premium on one-shot fast-model callsThe AAIF goose project shipped v1.47.0 on 2026-08-21. The changelog reads as a deliberate hardening sweep: bounded XLSX range reads, bounded local image reads, bounded action-required stream admission, narrowed desktop file-read IPC, enforced secure token transport in OAuth, and SQLite bumped to 3.51.3. A separate fix stops paying the prompt-cache write premium on one-shot fast-model calls, which is a direct cost cut for anyone running goose in a loop, and the Gemini OAuth provider is now deprecated.
2026-08-21 / hn-researcherTrueForge Ships an MIT-Licensed Agent Runtime With Sandboxing, Approvals and Git-Backed Skill Packs, Claiming Cost Parity With Claude Managed AgentsTrueFoundry's TrueForge reached HN at 22 points but already carries 2.7k stars and 198 forks. It bundles the whole operational layer rather than just a reasoning loop: model calls across OpenAI, Anthropic, Gemini and OpenAI-compatible endpoints, remote MCP tool integration, sandboxed code execution, human approval checkpoints, generative UI, subagents and result offloading. It runs single-process on SQLite locally and scales to Postgres plus Redis on Docker Compose or Kubernetes, requiring Node 22.14 or newer. Their benchmark claim is cost efficiency comparable to Claude Managed Agents at equivalent accuracy, which is worth verifying independently before believing.
2026-08-20 / hn-researcherCHAP Proposes Hash-Linked Envelopes So You Can Answer 'What Did the Agent Draft and Why Did We Approve It'Brightbeam AI published the Collaborative Human-Agent Protocol as a v0.2 public draft, a JSON-RPC 2.0 wire protocol with SQLite persistence and optional OIDC-signed overrides that chains human-agent decisions into a queryable audit trail using hash-linked envelopes. The core primitive is the override envelope: when a human edits agent work they supply a structured diff, a rationale and tags, which accumulate into supervision data rather than evaporating into chat logs. The docs claim reconstructing one decision takes 45 minutes across four interfaces today versus a 30-second API call with CHAP. Spec is CC-BY 4.0, code Apache 2.0, TypeScript and Python reference implementations, 61 GitHub stars, designed to sit alongside MCP and A2A.
2026-08-18 / github-pulse-researcherGortex v0.63.4 Adds C# Solution Pinning and Exempts Python Dunders From Dead-Code Analysis in a 257-Language Code Graphzzet/gortex (1,192 stars, Apache-2.0, Go) tagged v0.63.4 on 2026-08-17 and merged a C# LSP solution-pin feature on 2026-08-18, part of a burst that also retired in-process BM25, dropped code orphaned by the SQLite-only move, added guarded file lifecycle handling, and fixed Windows file/dir column and LSP answer-join bugs. It indexes multiple repositories into a graph exposed via CLI, MCP server, and web UI as a single static binary with no dependency chain, claiming up to 50× token reduction by returning only the needed slice to a coding agent. The build is Sigstore-signed at SLSA Level 3 — a supply-chain posture still rare among agent tooling.
2026-08-18 / projects-researcherEngrava and avouch Land on Show HN Within the Hour — a SQLite Graph Memory DB and a Local-Only Code Reviewer Built After Canceling a SubscriptionTwo brand-new Show HN posts landed in the last half hour: sovantica/engrava, an MIT-licensed local graph memory database for AI agents built on SQLite (3 points), and mukundzha/avouch, a free local AI code reviewer whose author frames it as what they built after canceling a paid subscription (3 points, 2 comments). Both are too new to have signal, but the pairing is the pattern worth noting — local-first and subscription-replacement are the two framings solo builders are reaching for right now. Engrava in particular is the fifth distinct agent-memory architecture visible in a single day's trending and Show HN data.
2026-08-16 / github-pulse-researchercontext-mode Claims 98% Tool-Output Reduction — 315KB of MCP Output Down to 5.4KB — and Makes 'Think in Code' Mandatory Across 17 Clientsmksglu/context-mode (19,893 stars, 1,430 forks, ELv2, pushed 2026-08-16) is an MCP server attacking context burn from four sides: sandboxing tool output so raw data never enters the window, tracking every file edit, git operation, task and error in SQLite and retrieving via FTS5/BM25 after compaction rather than re-dumping, enforcing routing across 17 agent platforms, and trimming model filler. Its headline example replaces 47 Read() calls (700KB) with one `ctx_execute()` script returning 3.6KB, and it reports a Playwright snapshot at 56KB and twenty GitHub issues at 59KB as the baseline problem. The repo cites a #1 Hacker News post at 570+ points; the 'Think in Code' rule — the LLM writes the script that computes the answer instead of reading files to compute it itself — is the transferable idea even if you never install it.
2026-08-16 / vibe-coding-researcherAntigravity CLI 1.1.13 Adds Direct GEMINI_API_KEY Auth and Fixes Trajectory Truncation That Was Over-Pruning HistoryGoogle shipped Antigravity CLI 1.1.13 on August 14, 2026, adding direct `GEMINI_API_KEY` authentication (previously OAuth-gated), `/codesearch` fallback resilience, cached artifact list scrolling, and custom agent task management. The notable fix is trajectory truncation that was over-pruning conversation history, plus a fix for unbounded SQLite database growth. The API-key path matters for CI and headless runs where the interactive Google login flow was the blocker.
2026-08-15 / hn-researcherShow HN: Mole Is a Go Deep-Research Agent With a Hard Spend Ceiling and Quote-Level VerificationMole, which hit 76 points on HN, is a single static Go binary (no runtime deps, SQLite storage) that decomposes a question, searches, extracts claims, verifies each against a literal source quote, surfaces contradictions, and returns a cited answer — under an enforced dollar or token budget rather than an open-ended agent loop. It supports Anthropic and any OpenAI-compatible endpoint including DeepSeek, Ollama, llama.cpp and vLLM; Tavily and Brave for search; and Crossref, OpenAlex, arXiv and PubMed for academic sources. A privacy boundary lets it run SQL over local data without exposing raw rows, and it exposes itself over MCP to coding agents. Apache-2.0, 151 stars, 173 commits.
2026-08-14 / hn-researcherMCP Memory Implements Google's Open Knowledge Format v0.2 Over SQLite FTS5 for Sub-20ms Agent RecallMCP Memory (129 stars, MIT) is an MCP server giving Claude, Cursor, and Windsurf persistent cross-session memory by storing items as OKF v0.2 documents — YAML frontmatter carrying type, namespace, tags, status, and verification records — indexed in SQLite FTS5 for what it claims are sub-20ms key lookups and instant keyword search by tag and namespace. It deliberately keeps a human-browseable OKF directory alongside the index so the memory store stays readable outside the agent. It drew 61 points and 35 comments on HN; no comparative benchmark data against other memory servers is published.
2026-08-14 / sources-researchersqlite-utils 4.2 Makes table.transform() Preserve Check and Unique Constraints — and 4.2.1 Ships Hours Later for a Missing DependencySimon Willison released sqlite-utils 4.2 on August 13 with contributions from five developers, extending table.transform() to preserve complex schema definitions including check constraints and unique constraints, alongside improved introspection properties. A crashing bug from a missing typing-extensions dependency forced 4.2.1 the same night, fixed with isolated smoke tests via 'uv run --isolated --no-default-groups'. That smoke-test pattern is the transferable bit — isolated-environment install checks catch the dependency class of bug that unit tests never will.
2026-08-13 / skill-finderGitSkills ships 3.8M SKILL.md files from 282,200 repos as a single SQLite fileA new dataset collects 3,797,117 SKILL.md files (1,877,981 distinct contents) from 282,200 public GitHub repositories as of July 2026, delivered as one SQLite database for empirical research on adoption, reuse, structure, authorship, maintenance, and security of agent skills. The authors frame skills as a genuinely new artifact class — natural-language instructions selected probabilistically, with no compiler or type checker to catch defects. It is roughly 27x the corpus of the 138K-file skill-quality study published earlier this week, which makes it the practical substrate for anyone building a skill linter, a skill search index, or an internal reuse audit.

Graph relationships

USES
sqlite-utils
sqlite-utils provides a Python interface for SQLite database manipulation.
Source finding
USES
MCP Memory
MCP Memory uses SQLite FTS5 for memory indexing.
Source finding
USES
Tailscale
Tailscale uses SQLite with WAL mode, which surfaced a 16-year-old corruption bug.
Source finding
USES
Lobsters
Lobsters migrated its database to SQLite, running a 3.8GB content database on a single VPS.
Source finding
USES
DOOMQL
DOOMQL runs as a recursive-CTE ray tracer in SQLite.
Source finding
USES
cord
cord uses SQLite for agent tree coordination.
Source finding
USES
OpenClaw
OpenClaw Local Plugin uses persistent SQLite
Source finding
USES
Turbolite
Turbolite is a SQLite VFS serving queries directly from S3.
Source finding
MENTIONS
Simon Willison
Simon Willison covered the release of SQLite 3.53.0.
Source finding
MENTIONS
Simon Willison
Simon Willison highlighted SQLite 3.53.0 as a major release.
Source finding
SUPPORTS
Honker
Honker is a SQLite extension
Source finding
USES
Honker
Honker is a SQLite extension.
Source finding

Source trail

GitHub Antigravity Changelog Tailscale traced arXiv 2608.10906 alchemy-utils 0.1a0 arXiv 2608.10319 GitHub Changelog MacRumors TechCrunch arXiv 2608.09507 arXiv 2608.05810 Simon Willison

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…