Skip to content

README: purpose, one-cycle narrative, actors, architecture diagram - #5

Merged
rbaker5 merged 1 commit into
mainfrom
docs/readme-architecture
Jul 14, 2026
Merged

rbaker5 merged 1 commit into
mainfrom
docs/readme-architecture

Conversation

@rbaker5

@rbaker5 rbaker5 commented Jul 14, 2026

Copy link
Copy Markdown
Contributor

Improves the README's explanation of purpose and functional design, then
caps it with an architecture diagram — supporting prose first, diagram
last, so the diagram's node labels reuse terms the prose establishes.

  • New "Why this exists" section: the problem (published findings decay
    silently) and the answer (the maintenance loop as the versioned
    artifact).
  • "One cycle, end to end": a linear narrative of
    sensor → descriptor → dedup → data-PR + provenance stub → interpretation
    → merge/decline → evolution narrative, with each canonical term bolded
    and linked to CONCEPT.md § Canonical terms.
  • Actors made explicit (engine under the GitHub App identity, gh-aw
    agents behind safe-outputs, managing author with sole merge authority),
    linked to CONCEPT.md § Roles.
  • Mermaid architecture diagram at the top of "How it works", grouped
    by trust layer, Phase-2 impact dashed as opt-in; prose caption as a
    fallback for non-mermaid renderers.
  • Fixes an engine-specificity error: the agent layer was described as
    "running Claude" — engine choice is per-instance (the sample runs
    Gemini); only sensor-repair is genuinely Claude-specific.

Please check the diagram renders correctly in the Files view before
merging. A matching README refresh for the sample instance is in
norabble/continuous-research-sample.

🤖 Generated with Claude Code

…re diagram

The README explained the two-layer split well but never said WHY the
project exists, never walked one cycle end to end, and left the three
actors implicit. Add a "Why this exists" section (findings decay; the
maintenance loop becomes the artifact), a "One cycle, end to end"
narrative that bolds each canonical term and links CONCEPT.md for the
vocabulary, an explicit actors list (engine under the App identity /
sandboxed agents / managing author), and a mermaid architecture diagram
grouping the flow by trust layer, with a prose caption as fallback.

Also correct an engine-specificity error: the agent layer was described
as "running Claude", but engine/model choice is per-instance (the sample
runs Gemini); only the optional sensor-repair integration is
Claude-specific and keeps saying so. "Adopting it" gains the one-line
you-bring / framework-provides hook.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@rbaker5
rbaker5 merged commit b485a5a into main Jul 14, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant