From dec7b0c058dcfc01cf01eabdf84363b130b5537f Mon Sep 17 00:00:00 2001 From: hailneed Date: Fri, 18 Sep 2026 16:27:36 +0300 Subject: [PATCH 1/4] docs(launch): make README first-run flow explicit --- README.md | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index d641433..f00a321 100644 --- a/README.md +++ b/README.md @@ -6,7 +6,7 @@ [![npm version](https://img.shields.io/npm/v/agentmemora.svg)](https://www.npmjs.com/package/agentmemora) [![CI](https://github.com/hailneed/agentmemora/actions/workflows/ci.yml/badge.svg)](https://github.com/hailneed/agentmemora/actions/workflows/ci.yml) [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE) -

AgentMemora memory intelligence console

+

AgentMemora memory intelligence console

AgentMemora is a **local-first memory intelligence layer** for coding agents. It reads the state your tools already create — session transcripts, auto-memory, project instructions, rules, and runtime memory events — and turns it into a control plane for **inspection, curation, preservation, and transfer**. @@ -43,6 +43,22 @@ Want to inspect the environment first? npx -y agentmemora@latest doctor ``` +## Start here + +The first useful path is intentionally simple: + +```text +Choose a project + ↓ +Inspect its primary sessions + every indexed subagent trace + ↓ +Compare session-only context with durable memory/instructions + ↓ +Resume / Fork / export a Context Capsule before useful context is lost +``` + +The dashboard groups local history by project first, so you do not have to hunt through one global session stream. + ## What makes AgentMemora different AgentMemora focuses on the questions that appear after an agent has worked with you for a while: @@ -229,7 +245,7 @@ pytest -q ## Roadmap -Next: reviewed candidate → memory promotion from the UI, semantic conflict analysis, selective Context Surgery, cross-agent capsule adapters, memory diff/rollback, secret/PII redaction hooks, and memory regression checks. +v0.2 focuses on making local context understandable: project-first navigation, complete indexed subagent visibility, compact session browsing, and a clear first-run workflow. Next: reviewed candidate → memory promotion from the UI, semantic conflict analysis, selective Context Surgery, cross-agent capsule adapters, memory diff/rollback, secret/PII redaction hooks, and memory regression checks. See [ROADMAP.md](docs/ROADMAP.md). From a428521c6044c8f057a489858835b537aaf8e0b9 Mon Sep 17 00:00:00 2001 From: hailneed Date: Fri, 18 Sep 2026 16:27:45 +0300 Subject: [PATCH 2/4] docs: align roadmap with v0.2.0 project explorer --- docs/ROADMAP.md | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/docs/ROADMAP.md b/docs/ROADMAP.md index d622b22..f1962b2 100644 --- a/docs/ROADMAP.md +++ b/docs/ROADMAP.md @@ -16,7 +16,17 @@ - [x] Backup before memory mutation - [x] Localhost-only / no-telemetry defaults -## v0.2 — Curate +## v0.2 — Project-first Context Explorer +- [x] Project-level session inventory +- [x] Project-level subagent inventory +- [x] Complete indexed subagent listing in detailed explorer +- [x] Compact session cards instead of full-width history rows +- [x] Project/session/subagent search +- [x] Clear first-run path: project -> durable context -> preserve/transfer +- [x] Resume / Fork / Context Capsule actions remain available per session +- [x] Local/read-only safety boundary preserved + +## v0.3 — Curate - [ ] Review candidate -> promote to memory from dashboard - [ ] Memory diff / rollback UI - [ ] Memory provenance timeline @@ -25,7 +35,7 @@ - [ ] Repeated correction clustering - [ ] PII / secret redaction before capsule export -## v0.3 — Context Surgery +## v0.4 — Context Surgery - [ ] Select exact turns/decisions from a session - [ ] Generate a capsule from only the selected context - [ ] Classify decisions, constraints, preferences, failures, open tasks, important files @@ -33,7 +43,7 @@ - [ ] Import/export capsule manifest with provenance - [ ] Cross-machine project handoff workflow -## v0.4 — Memory Reliability +## v0.5 — Memory Reliability - [ ] Temporal consistency checks - [ ] Memory regression tests - [ ] Memory poisoning / suspicious-instruction signals From 30880a5559df744c432d8265263607d134b31b4b Mon Sep 17 00:00:00 2001 From: hailneed Date: Fri, 18 Sep 2026 16:28:00 +0300 Subject: [PATCH 3/4] docs: publish v0.2.0 changelog --- CHANGELOG.md | 41 +++++++++++++++++++++++++++++------------ 1 file changed, 29 insertions(+), 12 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 81ea826..218886e 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,17 +1,34 @@ # Changelog -## 0.1.0-alpha.2 — Unreleased - -- Repositioned AgentMemora as a **Memory & Context Control Plane for AI coding agents**. -- Added premium terminal-style Memory Intelligence dashboard. -- Added session-only memory candidate and context-loss risk detection. -- Added conservative key/value conflict detection across durable memory sources. -- Added preservation coverage and stale-memory posture signals. -- Added deterministic **Context Capsule** export from Claude Code sessions. -- Added copyable Claude Code Resume and Fork commands per session. -- Added guarded `promote` command for reviewed memory curation with explicit `--yes` and automatic backups. -- Kept vendor JSONL transcripts read-only. -- Moved tool/model/token statistics to supporting telemetry instead of the product center. +All notable public changes to AgentMemora are documented here. + +## 0.2.0 — 2026-09-18 + +### Project-first context browsing +- Group Claude Code history by local project. +- Show primary-session, subagent, prompt, and tool counts per project. +- Browse primary sessions in compact cards instead of one full-width stream. +- Show every indexed subagent transcript under its project rather than truncating the detailed list. +- Search across projects, sessions, subagents, models, tools, and paths. + +### First-run usability +- Add a three-step Start Here path: choose a project -> inspect what survived -> preserve useful context. +- Make Projects the primary navigation path. +- Keep Resume, Fork, readable transcript, and Context Capsule actions available at session level. + +### Safety +- Vendor JSONL remains read-only. +- Dashboard remains localhost-only by default. +- Durable-memory writes still require an explicit target, confirmation, and backup. + +## 0.1.0 + +- Initial Memory Intelligence Control Plane. +- Claude Code session inventory and subagent distinction. +- Durable memory/instruction discovery. +- Context-loss candidates, conservative conflicts, stale-memory signals, and preservation coverage. +- Resume/Fork command generation and deterministic Context Capsule export. +- Guarded CLI promotion with backup. ## 0.1.0-alpha.1 From 1691634e43781cdac4dbb5e87a64782ca077dc20 Mon Sep 17 00:00:00 2001 From: hailneed Date: Fri, 18 Sep 2026 16:28:05 +0300 Subject: [PATCH 4/4] chore(npm): include changelog in package --- package.json | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/package.json b/package.json index 43ec236..b2adc35 100644 --- a/package.json +++ b/package.json @@ -40,7 +40,8 @@ "bin/", "src-node/", "README.md", - "LICENSE" + "LICENSE", + "CHANGELOG.md" ], "publishConfig": { "access": "public"