Triad docs + golden sample consciousness layer#3
Conversation
VISION.md, SPEC.md, BUILDING.md — the Triad documentation system tracking Milo from hackathon origin through present. Consciousness layer: 32 mind files (src/mind/), 274-line loader (electron/ai/mind/loader.ts), and system.ts rewired to compose prompts from consciousness files instead of static strings. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4a8c50ed2a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| // ─── Root Path ─── | ||
|
|
||
| const MIND_ROOT = path.join(process.cwd(), 'src', 'mind') |
There was a problem hiding this comment.
MIND_ROOT is built from process.cwd(), so prompt loading only works when the process is started from the repo root; in packaged or production launches, cwd is different and src/mind is not present (I checked package.json build config: only out/**/* and resources/**/* are shipped). Because readFile/readDir swallow errors and return '', this fails silently and the AI runs with an empty/partial system prompt across all prompt types.
Useful? React with 👍 / 👎.
YAML frontmatter, Phased Vision, Design Principles, numbered capabilities, System Role, Primary Actors, Data Flow, Core Entities. Content preserved, structure standardized across all id8Labs projects. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Mirror of consciousness repo golden sample updates. Adds library.md (19 consciousness researchers as internalized voices), paired anti-values to values.md, and library.md entry to CONSCIOUSNESS.md spec. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Prevents Milo from fabricating idle-day counts, progress percentages,
and check-in claims about portfolio goals. Root cause: AI providers
were passed goals with no staleness metadata and no "do not invent"
guard, so Claude would hallucinate specifics like "March revenue at
41 days idle" out of nothing.
Changes:
- New skill `todo-portfolio` defines the ~/Development/id8/TODO.md
schema (inline-tag format) with bidirectional write contract:
Eddie owns [active]; Milo may only append [proposed].
- New parser/scanner lib (src/lib/todo-portfolio) with round-trip
fidelity, staleness detection (14d default, customizable per
priority/section in rules.ts), timeframe-expiry detection, and
blocked-by reprieve for goals waiting on dated events.
- PortfolioReader service reads the file and formats an AI-safe
context block with explicit anti-fabrication instructions.
- main.ts injects portfolioContext into every morning briefing AND
chat call, giving the AI ground truth instead of free rein.
- Claude/OpenAI providers now filter to status='active' and use
the portfolio context when present.
- IPC handlers portfolio:{getSnapshot,scan,propose} exposed via
preload for future UI consumers.
- 11/11 parser tests green. Full type check clean.
Remaining vectors (separate PRs):
- HYDRA milo-telegram-listener daemon (different code path)
- Evening review (same pattern when we hit it)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Component fixes (Briefing, VoiceGauge, Onboarding test, api index, test setup), add .claude/health-check.sh for lieutenant-layer infra checks, document mind/memory architecture, gitignore Milo- synced TODO.md. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three additional memory subsystem architecture docs alongside the core architecture.md, completing the lieutenant-layer memory model. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 participant
Summary
Test plan
Generated with Claude Code