Entity trail

datasette-llm

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-02 / sources-researcherMCP 2.0 Went Stateless on July 28 — One HTTP Request, No Mcp-Session-Id, and Willison Immediately Shipped Three ClientsThe July 28, 2026 MCP specification revision replaces session-based transport with 'stateless, self-contained requests' and 'per-request capability negotiation'; the old two-round-trip dance (initialize → receive Mcp-Session-Id → call tool) collapses into a single HTTP request carrying protocol version and method in headers. Simon Willison, previously lukewarm on MCP, wrote on July 31 that this 'recaptured my interest' and shipped three tools in a day: mcp-explorer (uvx-installable CLI), datasette-mcp (adds a /-/mcp endpoint with list_databases, get_database_schema, and read-only execute_sql), and llm-mcp-client. The spec also formalizes three opt-in extensions — Tasks (async long-running ops with durable handles), Skills over MCP, and MCP Apps (inline charts/forms/video) — meaning servers no longer need sticky routing to scale horizontally.
2026-06-19 / thought-leaders-researcherSimon Willison Ships 'Datasette Apps' — 'Claude Artifacts Reimagined for Datasette' With a JSON API to a Full Relational DatabaseOn June 18 Simon Willison launched datasette-apps (0.1a2), a plugin that hosts self-contained HTML+JavaScript applications inside a sandboxed iframe in Datasette, where they can run read-only (or, via stored queries, write) SQL against your data. He frames it explicitly as 'Claude Artifacts reimagined for Datasette' — artifacts-style apps but backed by a JSON API to a real relational database rather than ephemeral state. The plugin has no LLM dependency, but the create-app form ships a copyable prompt pre-loaded with the selected database schema so a model like Claude, ChatGPT, or Gemini can one-shot a working data app; he also released datasette-acl 0.6a0 generalizing permissions into a resource-sharing system. For builders, it's a concrete pattern for turning LLM-generated artifacts into persistent, data-backed internal tools.
2026-06-18 / sources-researcherSimon Willison: Can DuckDB Run Untrusted SQL as Safely as Datasette Runs SQLite?Willison's research post concludes DuckDB can match or exceed SQLite's safety for untrusted SQL, but only with a hardened config (enable_external_access=false, lock_configuration=true) plus a watchdog thread, since DuckDB lacks SQLite's opcode-based query timeouts. He ships a safe_duckdb.py helper and a datasette_duckdb.py prototype that serves a DuckDB file securely through Datasette's web interface. Directly actionable for anyone exposing analytical SQL to users or LLM-generated queries over Parquet/columnar data.
2026-06-10 / sources-researcherSimon Willison's LLM 0.32a3 Adds Tool-Call Pause/Resume Primitives — Built Almost Entirely by Fable 5Driven by a human-in-the-loop ask_user() feature for Datasette Agent, LLM 0.32a3 (released 2026-06-09) adds a PauseChain exception to cleanly pause a tool chain for approval, guarantees a unique tool_call_id per call (synthesizing tc_-prefixed ULIDs when providers omit them), and lets chains resume from a message history ending in unresolved tool calls. Willison says Fable wrote nearly all of it — unraveling hacks into supported LLM features and producing the API design, tests, and docs across both LLM and Datasette Agent in what felt like several days of work compressed into one. These are directly useful primitives for builders adding human-approval gates to agentic tool execution.
2026-05-24 / vibe-coding-researcherSimon Willison Releases Datasette Agent 0.1a3 — AI Assistant for SQLite DatabasesSimon Willison released Datasette Agent 0.1a3 on May 21, 2026, an extensible AI assistant plugin for Datasette that provides a conversational interface for querying and visualizing SQLite data. Built on the LLM Python library, it supports hundreds of tool-calling models from OpenAI, Anthropic, Google, and local open-weight models. Additional plugins add charting (datasette-agent-charts) and sprite generation. A live demo is available at agent.datasette.io.
2026-05-16 / sources-researcherSimon Willison: datasette-agent 0.1a2 — Agentic LLM Capabilities Added to Datasette FrameworkSimon Willison released datasette-agent 0.1a2, adding agentic LLM capabilities to the Datasette ecosystem. Part of a broader May 2026 push from Willison to make Datasette a complete LLM application platform — alongside datasette-llm, datasette-llm-accountant, and datasette-llm-limits, this provides tool-using agent functionality within the data exploration framework.
2026-05-16 / sources-researcherSimon Willison: datasette-llm-limits — Per-User LLM Spending Controls for Multi-Tenant Datasette ApplicationsSimon Willison released datasette-llm-limits 0.1a0, a plugin that enforces monetary spending caps on LLM API usage within Datasette instances. Works with datasette-llm (core LLM functionality) and datasette-llm-accountant (cost tracking) to provide granular control: per-user daily limits, rolling 24-hour windows, and maximum spend amounts in USD. Solves the 'unexpected LLM bill' problem for multi-tenant deployments.
2026-05-14 / thought-leaders-researcherSimon Willison Releases datasette 1.0a29 — Approaching 1.0 MilestoneSimon Willison released datasette 1.0a29 on May 12, continuing the push toward a stable 1.0 release of the popular data exploration tool. Combined with the new Datasette blog and the llm 0.32a2 release on the same day, this represents an unusual burst of shipping activity from one of the developer community's most prolific open-source maintainers. The simultaneous releases suggest Willison is using AI tools to accelerate his own release cadence.
2026-05-13 / sources-researcherSimon Willison Releases llm 0.32a2 — CLI Tool for Accessing LLMs Gets Major Alpha UpdateSimon Willison released llm 0.32a2 on May 12, the second alpha of a major version bump for his popular open-source LLM CLI tool that provides a unified command-line interface for interacting with language models. The 0.32 series introduces consequential changes to the tool's architecture. Alongside this, datasette 1.0a29 was also released on the same day. For developers who use llm as part of their AI scripting toolkit, this alpha is worth testing to prepare for breaking changes.
2026-05-06 / sources-researcherSimon Willison: datasette-llm 0.1a7 — Model-Level Default Options for Datasette's LLM Plugin EcosystemDatasette-llm 0.1a7 introduces a mechanism for configuring default options (like temperature) per model, so dependent plugins like datasette-enrichments-llm inherit standardized configurations without per-operation setup. This is a building block in Datasette's evolving LLM infrastructure — reducing friction for plugin authors and creating more consistent behavior across the ecosystem.
2026-05-05 / sources-researcherSimon Willison: April 2026 Newsletter — Independent Analysis of Opus 4.7, GPT-5.5, and the State of AI EngineeringWillison published his sponsors-only April 2026 newsletter covering Opus 4.7, GPT-5.5, and the broader state of AI engineering tools. As one of the most trusted independent voices in the AI developer space, his monthly synthesis typically surfaces under-discussed angles and practical implications that don't appear in mainstream coverage. The newsletter also covers developments in his datasette/LLM ecosystem.
2026-04-16 / sources-researcherSimon Willison: datasette-ports 0.3 Released — Datasette's Network Port Management PluginSimon Willison released datasette-ports 0.3, a plugin for managing network ports in Datasette, the open-source multi-tool for exploring and publishing data. This continues Willison's steady cadence of Datasette ecosystem improvements, alongside his LLM tool which reached version 0.30 on April 7. For builders running local data exploration workflows, Datasette remains one of the most practical tools for quickly exposing SQLite databases via a web interface.

Graph relationships

RELEASED
Simon Willison
Simon Willison released datasette-llm 0.1a1 enabling purpose-based model routing for plugins
Source finding
SUPPORTS
Datasette
datasette-llm extends Datasette with LLM-powered enrichment capabilities.
Source finding
BUILT BY
Simon Willison
Simon Willison released datasette-llm 0.1a6 with actor tracking.
Source finding
SUPPORTS
Datasette
datasette-llm is a plugin extending Datasette's LLM ecosystem
Source finding
RELEASED
Simon Willison
Simon Willison released datasette-llm 0.1a7 with model-level default options
Source finding
USES
datasette-extract
datasette-extract uses datasette-llm for model configuration.
Source finding
RELEASED
Simon Willison -> datasette-llm
Simon Willison released datasette-llm 0.1a1 enabling purpose-based model routing for plugins
Source finding
RELEASED
Simon Willison -> datasette-llm
Simon Willison released datasette-llm 0.1a7 with model-level default options
Source finding
USES
datasette-extract -> datasette-llm
datasette-extract uses datasette-llm for model configuration.
Source finding

Source trail

Simon Willison (with the modelcontextprotocol.io 2026-07-28 spec as primary)simonwillison.net Simon Willison Simon Willison / llm.datasette.io Simon Willison simonwillison.net simonwillison.net simonwillison.net Simon Willison Simon Willison Simon Willison Simon Willison

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…