[magnus919]

What changes

The 1.0.0 documentation did not clearly distinguish the original demonstration from the live research workflow. This update brings the architecture diagram, ADR index, README, developer instructions, API/CLI/MCP guides, and migration and recovery instructions into line with the completed 1.0.1 fixes.

The guides explain how retained research becomes session evidence, how explicit follow-ups run, what survives a restart, and how to migrate the two vector indexes independently. They also state the limits: uncertain external calls are not silently repeated, time-based evidence checks do not retrieve archived pages, and controlled integration tests do not prove better live-model answers.

The development instructions now use test files that exist and explain that production containers do not ship the test suite. Historical experiment records remain historical.

Validation

Documentation surface checks pass against routes and configuration. CLI coverage checks pass for 52 endpoints (17 exemptions); MCP coverage checks pass for 64 endpoints (59 tools and five exemptions). All 175 relative file links in the changed Markdown resolve. An independent source review found one unsafe configuration-copy instruction, which is corrected.

Implementation acceptance is recorded in the cleanup plan with its tested revisions and CI links. Deployment and release-image verification remain separate requirements; these documentation checks do not establish live-model answer quality.

Release-As: 1.0.1

Implementation base: merged PR #674 at 3f1406970c900ab999b04bb07009c2c16bb2eeeb; both integration suites passed 2,798 tests, with 162 skips and 17 deselections recorded separately. ADR-0070 is accepted on that evidence.

[factory-droid]

Droid encountered an error —— View job

Droid Exec exited with code 1:
Tool "Edit" skipped (Nous-GPT-6-Luna uses ApplyPatch instead).
Tool "Create" skipped (Nous-GPT-6-Luna uses ApplyPatch instead).

---