Clean up Memory Recall repository and product identity

.gitignore

@@ -3,6 +3,7 @@ node_modules/
 .local/
 coverage/
 .playwright-cli/
+.scratch/
 .worktrees/
 output/playwright/
 context-packs/

AGENTS.md

@@ -2,7 +2,7 @@
 
 ## Mission
 
-Develop Open Agent Fabric into the complete local-first, model-neutral tool for inspectable agents. Optimize for correctness, evidence, portability, and safe evolution, not feature count or autonomous behavior.
+Develop Memory Recall into the local-first, model-neutral repository memory and context tool for coding agents. Optimize for correctness, evidence, portability, and safe evolution, not feature count or autonomous behavior.
 
 ## Read order before changing code
 
@@ -67,7 +67,7 @@ npm run ci
 
 ### Issue tracker
 
-Issues are tracked in GitHub Issues for `rebel0789/open-agent-fabric`; local markdown under `.scratch/` may be used for offline or scratch drafts. External PRs are a triage surface. See `docs/agents/issue-tracker.md`.
+Issues are tracked in GitHub Issues for `rebel0789/Memory-Recall`; local markdown under `.scratch/` may be used for offline or scratch drafts. External PRs are a triage surface. See `docs/agents/issue-tracker.md`.
 
 ### Triage labels
 
@@ -81,7 +81,7 @@ This repo uses a single-context layout: root `CONTEXT.md` when present, plus arc
 
 1. **Canonical events:** state-changing operations emit append-only events.
 2. **Context manifests:** every model call records selected and excluded context with reason codes.
-3. **Own the tool boundary:** OAF ships the default runtime, context, memory, source-graph, tool, hook, UI, and CLI experience. Integrations may accelerate inputs or scale a backend, but they never own canonical state or authority.
+3. **Own the tool boundary:** Memory Recall ships the default runtime, context, memory, source-graph, tool, hook, UI, and CLI experience. Integrations may accelerate inputs or scale a backend, but they never own canonical state or authority.
 4. **Deterministic authority:** permissions, budgets, approvals, retries, and transitions are code—not model decisions.
 5. **No silent memory overwrite:** use lifecycle states and `supersedes` links.
 6. **Observations are not interpretations:** keep collected data separate from model analysis.

ASSIGN_TO_AGENT.md

@@ -1,10 +1,10 @@
 # Assignment Brief for the Development Agent
 
-You are receiving an agent-ready development kit for **Open Agent Fabric 0.2.0-dev**. Evolve it through small verified changes. Do not rebuild the architecture from scratch and never claim that specified components already exist.
+You are working on **Memory Recall 1.1.0**, currently an unreleased release candidate. Evolve it through small verified changes. Do not rebuild the architecture from scratch and never claim that specified capabilities already exist.
 
 ## Mission
 
-Build the complete local-first reliability and context tool for agents whose state, actions, evidence, memory, source understanding, workflows, permissions, and improvements are portable and inspectable.
+Build the local-first repository memory and context tool that gives coding agents reviewed facts, source-backed context, changed-file impact, and inspectable handoffs without making a hosted service the system of record.
 
 ## First session
 
@@ -40,7 +40,7 @@ or external adapters unless the active task explicitly requires it.
 
 ## Product differentiator
 
-Open Agent Fabric does not win by bundling repositories or outsourcing its core product behavior. It owns:
+Memory Recall does not win by bundling repositories or outsourcing its core product behavior. It owns:
 
 - intentional context selection and context manifests;
 - source graph, retrieval, compaction, and code intelligence for the default local experience;
@@ -83,4 +83,4 @@ Safest next task:
 
 ## Product success
 
-A contributor can run one useful local workflow, inspect exactly what context and evidence it used, replay it without side effects, replace a model or backend without losing portable state, and understand every consequential action before it occurs.
+A contributor can install Memory Recall in another repository, produce a useful read-only map and handoff, inspect exactly which context and evidence it selected, and review every memory proposal before it becomes active.

CHANGELOG.md

@@ -1,6 +1,6 @@
 # Changelog
 
-## [1.1.0] — 2026-07-11
+## [Unreleased] — 1.1.0 release candidate
 
 No breaking CLI migration is required. Local-only defaults, proposal-gated memory, and read-only MCP remain unchanged.
 
@@ -13,6 +13,7 @@ No breaking CLI migration is required. Local-only defaults, proposal-gated memor
 
 ### Changed
 
+- Canonical repository guidance, citation metadata, support links, project status, and generated manifest now use Memory Recall; legacy `oaf` commands, URIs, task IDs, and internal package scopes remain compatibility identifiers.
 - Replaced the fifteen-item dashboard navigation with a five-destination workbench, separated local workspace setup from Overview, and made Overview choose one deterministic scan, review, or handoff action from current local state.
 - Overview now hydrates bounded multi-file impact from current read-only git detection, preserves omitted-change evidence, reports degraded detection without claiming a clean tree, surfaces blocked handoffs, and exposes repository search with stable Map history and explicit failure states.
 - The repository bar now documents the one-repository-per-server boundary, and tablet navigation collapses to a semantic 72 px icon rail with accessible labels.
@@ -29,7 +30,27 @@ No breaking CLI migration is required. Local-only defaults, proposal-gated memor
 
 ### Internal
 
-- Release gates cover 655 Node tests, 182 protocol fixtures, 144 evaluation assertions, installed-package smoke, browser smoke, package-content verification, and release-evidence drift checks.
+- Release gates cover 659 Node tests, 182 protocol fixtures, 144 evaluation assertions, installed-package smoke, browser smoke, package-content verification, and release-evidence drift checks.
+
+## [1.0.5] — 2026-07-09
+
+### Added
+
+- Bounded source-graph commands, previews, protocol examples, measurements, and focused validation coverage.
+
+### Changed
+
+- GitHub release metadata now points to the exact commit recorded by the published `memory-recall@1.0.5` package.
+
+## [1.0.4] — 2026-07-08
+
+### Added
+
+- Model-free LoCoMo-shaped retrieval benchmark with a small checked-in smoke fixture.
+
+### Changed
+
+- Public benchmark copy distinguishes retrieval coverage from official generative QA scoring.
 
 ## [1.0.0] — 2026-07-08
 

CITATION.cff

@@ -1,10 +1,8 @@
 cff-version: 1.2.0
-message: "If you use Open Agent Fabric, cite the project and release."
-title: "Open Agent Fabric"
-version: "0.2.0-dev"
-date-released: 2026-06-19
+message: "If you use Memory Recall, cite the project."
+title: "Memory Recall"
 license: Apache-2.0
-repository-code: "https://github.com/rebel0789/open-agent-fabric"
+repository-code: "https://github.com/rebel0789/Memory-Recall"
 authors:
   - family-names: "Contributors"
-    given-names: "Open Agent Fabric"
+    given-names: "Memory Recall"

CONTEXT.md

@@ -1,6 +1,6 @@
-# Open Agent Fabric
+# Memory Recall
 
-Open Agent Fabric is a local-first context and reliability system for coding agents. This glossary defines the product language agents should use when planning, implementing, or reviewing work in this repository.
+Memory Recall is a local-first repository memory and context system for coding agents. This glossary defines the product language agents should use when planning, implementing, or reviewing work in this repository. Legacy `oaf` commands, URIs, package scopes, and task IDs remain compatibility or internal identifiers.
 
 ## Language
 
@@ -9,27 +9,27 @@ A local control layer that gives coding agents selected context, memory proposal
 _Avoid_: Agent dashboard, hosted memory SaaS, universal automation platform, codebase-memory clone
 
 **Context Quality Ranking**:
-The product priority order for context work: correct and safe context first, token efficiency second, speed third, and harness breadth fourth. OAF should only claim optimization when required evidence and safety are preserved.
+The product priority order for context work: correct and safe context first, token efficiency second, speed third, and harness breadth fourth. Memory Recall should only claim optimization when required evidence and safety are preserved.
 _Avoid_: Token savings at any cost, fastest response wins, broad harness support as proof of quality
 
 **Safe Context Layer**:
-The first public beta promise: OAF runs locally and gives coding agents safe, token-efficient repository and memory context through context packs, read-only MCP resources, and proposal-gated memory writes.
+The first public beta promise: Memory Recall runs locally and gives coding agents safe, token-efficient repository and memory context through context packs, read-only MCP resources, and proposal-gated memory writes.
 _Avoid_: Hidden transcript upload, automatic cloud memory, invisible context injection, write-enabled MCP by default
 
 **Delivery Budget**:
-The estimated token and byte cost of the locator-only handoff that OAF actually gives another agent. It is tracked separately from the source budget, which estimates the full underlying context records that were scanned and selected.
+The estimated token and byte cost of the locator-only handoff that Memory Recall actually gives another agent. It is tracked separately from the source budget, which estimates the full underlying context records that were scanned and selected.
 _Avoid_: Treating raw source-body token counts as the cost of a sanitized handoff, claiming token savings without a delivered-context measurement
 
 **Proposal-Gated Memory Import**:
-The rule that imported harness context or transcript-derived learnings can become memory proposals, but cannot become active memory until the user reviews and accepts them. Users choose which Codex, Claude Code, Cursor, or other harness sources OAF may inspect.
+The rule that imported harness context or transcript-derived learnings can become memory proposals, but cannot become active memory until the user reviews and accepts them. Users choose which Codex, Claude Code, Cursor, or other harness sources Memory Recall may inspect.
 _Avoid_: Automatic active memory, silent transcript capture, default import from hidden app state
 
 **Read-Only MCP Surface**:
-The first MCP exposure for OAF: resources that let harnesses inspect status, context packs, manifests, source graph results, memory proposals, accepted-memory summaries, benchmark reports, and handoff bundles without mutating OAF state.
+The first MCP exposure for Memory Recall: resources that let harnesses inspect status, context packs, manifests, source graph results, memory proposals, accepted-memory summaries, benchmark reports, and handoff bundles without mutating Memory Recall state.
 _Avoid_: Write-enabled MCP by default, memory write tools before exact grants, MCP resources with hidden side effects
 
 **Native Source Graph**:
-A local derived index over normalized OAF source records, symbols, imports, calls, routes, tests, configuration, evidence, memory proposals, context manifests, and benchmark cases. It should support graph search, structural filters, call/data-flow traces, architecture summaries, diff impact, and bounded context selection while remaining an index, not canonical state.
+A local derived index over normalized repository records, symbols, imports, calls, routes, tests, configuration, evidence, memory proposals, context manifests, and benchmark cases. It should support graph search, structural filters, call/data-flow traces, architecture summaries, diff impact, and bounded context selection while remaining an index, not canonical state.
 Current implementation exposes a read-only JS/TS graph preview for search, call trace, and diff impact through CLI/API. It is not a persisted graph database and does not create memory.
 Context packs also include compact source graph hints as locators and fingerprints only; they do not embed raw code slices.
-_Avoid_: Source graph as canonical memory, graph database requirement, silent adapter enablement, mutating graph tools by default, token-saving claims without OAF benchmarks
+_Avoid_: Source graph as canonical memory, graph database requirement, silent adapter enablement, mutating graph tools by default, token-saving claims without Memory Recall benchmarks

GOVERNANCE.md

@@ -1,6 +1,6 @@
 # Governance
 
-Open Agent Fabric begins with maintainer-led governance and must evolve toward neutral stewardship before 1.0.
+Memory Recall uses maintainer-led governance and should evolve toward neutral stewardship as the contributor base grows.
 
 - Major protocol and architecture changes use public RFCs and ADRs.
 - Maintainers merge by evidence, compatibility, security, and user value.

MAINTAINERS.md

@@ -1,14 +1,14 @@
 # Maintainers
 
-Open Agent Fabric is currently maintainer-led by the repository owner:
+Memory Recall is currently maintainer-led by the repository owner:
 
 - GitHub owner: `@rebel0789`
-- Repository: <https://github.com/rebel0789/open-agent-fabric>
+- Repository: <https://github.com/rebel0789/Memory-Recall>
 
 ## Security and Conduct
 
 Use private GitHub Security Advisories for suspected vulnerabilities:
-<https://github.com/rebel0789/open-agent-fabric/security/advisories/new>
+<https://github.com/rebel0789/Memory-Recall/security/advisories/new>
 
 Use GitHub Discussions and Issues for community support and conduct escalation.
 Do not publish private security or conduct reports in public issues.

NOTICE

@@ -1,4 +1,4 @@
-Open Agent Fabric
-Copyright 2026 Open Agent Fabric contributors
+Memory Recall
+Copyright 2026 Memory Recall contributors
 
 This product includes no vendored source from the optional upstream projects listed in THIRD_PARTY.md.

PROJECT_STATUS.json

@@ -1,6 +1,6 @@
 {
   "schemaVersion": "1.0.0",
-  "project": "open-agent-fabric",
+  "project": "memory-recall",
   "release": "1.1.0",
   "phase": "1.1-release-candidate-readiness",
   "nextTask": null,
@@ -1026,7 +1026,7 @@
     }
   ],
   "qualitySnapshot": {
-    "asOf": "2026-07-11",
+    "asOf": "2026-07-15",
     "commands": [
       "npm ci --ignore-scripts --no-audit --no-fund",
       "npm run local:run",
@@ -1048,7 +1048,7 @@
       "npm run consumer:smoke",
       "npm run consumer:browser-smoke"
     ],
-    "testsExpected": 655,
+    "testsExpected": 659,
     "protocolFixturesExpected": 182,
     "evaluationsExpected": 144,
     "note": "Counts are a dated snapshot; command results and HANDOFF_VERIFICATION.json are authoritative."