Entity trail

transcripts-mcp

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-08 / vibe-coding-researchertranscripts-mcp lets one coding tool search the session history of the other twoThe server reads Cursor, Claude Code and Codex session transcripts already on disk and exposes them over MCP with fuzzy, plain-text and regex search, a local full-text index and optional semantic ranking, plus session browsing filtered by project or date. No index build is required for basic search. It returns transcript text for the current client to read, so it recovers reasoning and prior implementations across tools rather than restoring a previous session's model state; requires Node 24, created 2026-09-05.
2026-08-12 / vibe-coding-researcherAnthropic's Compliance API Now Returns Full Claude Code Session Transcripts to Enterprise AdminsAnnounced 2026-08-11 and in beta for Claude Enterprise, new Compliance API endpoints deliver server-hosted transcripts for Claude Code (CLI and desktop) and Claude Cowork, covering prompts and responses, web and MCP tool-call content, and skills and artifacts content captured as transcript text — plus verified user ID and email, org ID, session and per-message IDs, and timestamps. Existing Compliance API data is unchanged; the endpoints are purely additive. Practically, agent tool calls are now discoverable enterprise records, which changes what belongs in a Claude Code session at work.
2026-08-09 / github-pulse-researcherdeja-vu Indexes the Coding-Agent Session Logs Already Sitting on Your Diskvshulcz/deja-vu (612 stars, created July 14) takes a different angle on agent memory than the retain/recall systems: it indexes the session transcripts your coding agents already wrote to disk, turning months of prior history into a searchable store on the premise that 'your agents already solved this.' It ships as an MCP server, so any harness that speaks MCP can query it without adopting a new memory framework or re-ingesting anything. For anyone running agents daily, this is the cheapest possible memory layer — the corpus already exists locally and costs nothing to produce.
2026-08-09 / sources-researcherAnthropic's 'Running Auto Mode in Production' Gives the Operational Playbook: 9x Longer Between Interruptions, ~10% Denial Rate at Gusto, Interactive Mode for TerraformThe companion post to the auto mode default reports that across all Claude Code usage, sessions now run 9x longer between interruptions than under the previous permission default, and that at Gusto roughly 10% of session transcripts contained at least one auto-mode denial — the classifier is firing without stalling legitimate work. The recommended pattern is layered, not classifier-only: Gusto routes MCP traffic through a governed proxy with tool guards and prompt inspection before auto mode ever sees an action, Nuro disables recursive deletes outright in settings, and Garner Health blocks Slack and email sending on the user's behalf. Teams still flip back to interactive mode for production infrastructure — Terraform, AWS, direct POSTs against live APIs — and Gusto's Evan Magnussen names telemetry as the precondition: without it, 'that would be very dangerous.'
2026-08-07 / github-pulse-researcherCambium Proposes a Governance Standard for Knowledge Corpora That LLMs Maintain ThemselvesKimGLee/Cambium (created 2026-08-04, 204 stars, pushed 2026-08-07) is a governance standard plus reference toolset for corpora that agents write to and edit over time — the provenance, review, and drift problem nobody solves when an agent's memory store becomes the source of truth. It appears alongside bybit-exchange/kaas (88 stars, created 2026-07-29), an MIT knowledge-base compiler that turns notes and transcripts into a queryable Markdown wiki with MCP access and deliberately no embeddings. Both point at the same shift: agent memory is becoming a governed artifact, not a vector blob.
2026-08-02 / github-pulse-researcherCrypto Exchange Bybit Open-Sources an LLM Knowledge Compiler That Refuses to Use Embeddingsbybit-exchange/kaas appeared under the exchange's official GitHub org on 2026-07-29 and reached 70 stars with commits as recent as August 2. Rather than chunking and embedding, it compiles raw notes, docs and transcripts through a four-phase Extract, Classify, Write, Index pipeline into human-readable Markdown articles you can edit and version-control, with retrieval done LLM-iteratively from a master index down to page selection and article context. The stack splits a Go backend for the REST API, worker pool and MCP endpoint from a Python daemon running the LLM pipeline, with React on top and SQLite by default. It exposes a single ask(query, paths?, model?) tool over MCP with cited answers, in both stdio and HTTP transports. MIT, self-hosted. No performance or retrieval-quality numbers are published, which is the gap for anyone comparing it against a vector store.
2026-05-03 / skill-finderClaude Code /less-permission-prompts: Auto-Generate Allowlists from Session Transcripts to Reduce Permission FatigueA new built-in Claude Code skill that scans your recent session transcripts for common read-only Bash and MCP tool calls, then proposes a prioritized allowlist for your settings. Addresses the #1 friction point in agentic coding: constant permission prompts breaking flow state. The skill identifies patterns like repeated grep/find/git-log calls and suggests adding them to auto-allow. Invoke with /less-permission-prompts after a few sessions to build up your personalized allowlist.
2026-05-02 / vibe-coding-researcherClaude Code 2.1.126: Project Purge Command, Gateway Model Picker, and CLI UltrareviewReleased May 1, Claude Code 2.1.126 adds `claude project purge [path]` to delete all project state (transcripts, tasks, file history) with --dry-run and --interactive flags, the `/model` picker now lists models from Anthropic-compatible gateways via `/v1/models`, and `claude ultrareview [target]` enables non-interactive multi-agent code review from CI/CD scripts. Also includes OAuth paste-code fallback, MCP auto-retry on startup errors, and Windows PowerShell as default shell when Git Bash is absent.
2026-04-06 / github-pulse-researchertobi/qmd: On-Device Doc Search Engine with BM25 + Vector + LLM Re-Ranking and MCP Server — 18.5K StarsQMD is a local-first search engine that indexes markdown docs, meeting transcripts, and knowledge bases using a triple-layer approach: BM25 full-text, vector semantic search via node-llama-cpp with GGUF models, and LLM re-ranking — all running on-device. It ships with an MCP server for direct integration with AI agents and Claude Desktop. For anyone building agent-accessible documentation systems, qmd solves the local retrieval problem without any cloud dependency, and the MCP transport makes it immediately usable by coding agents.

Graph relationships

USES
Codex
transcripts-mcp server reads session transcripts from Codex
Source finding
USES
Claude Code
transcripts-mcp server reads session transcripts from Claude Code
Source finding
USES
Cursor
transcripts-mcp server reads session transcripts from Cursor
Source finding

Source trail

GitHub Anthropic GitHub Anthropic / Claude blog GitHub GitHub Claude Code Docs — What's New Claude Code Docs GitHub

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…