English · العربية · Español · Français · 日本語 · 한국어 · Tiếng Việt · 中文 (简体) · 中文（繁體） · Deutsch · Русский

Documentation-first repository scaffold for maintaining one canonical English README and synchronized multilingual documentation, guided by three operating principles: sear creation tools, self-healing tools, and chain of prompt tools.

AgInTi currently functions as a README lifecycle and localization pipeline, not as a runtime application. The root README.md is the canonical source, and localized variants in i18n/ are synchronized from that canonical structure.

The project philosophy is operational rather than decorative. Each README update is expected to satisfy all three principles:

1. Sear creation tools: intentionally sharp creation workflows that generate high-signal documentation from constrained repository evidence.
2. Self-healing tools: repair-oriented mechanisms that remove drift, duplication, and structural inconsistency.
3. Chain of prompt tools: staged, traceable prompt flows that preserve context-to-output lineage across pipeline runs.

This repository preserves meaningful historical content through incremental edits while retaining critical links, commands, and support metadata.

• README-first documentation strategy with a canonical root document.
• Multilingual synchronization across 10 i18n README variants.
• Pipeline-driven authoring via .auto-readme-work/<run-id>/ artifacts.
• Single-banner and single-support-panel invariants to prevent duplicate visual blocks.
• Incremental-update discipline that preserves substantive technical history.

AgInTi/
├── README.md
├── i18n/
│   ├── README.ar.md
│   ├── README.de.md
│   ├── README.es.md
│   ├── README.fr.md
│   ├── README.ja.md
│   ├── README.ko.md
│   ├── README.ru.md
│   ├── README.vi.md
│   ├── README.zh-Hans.md
│   └── README.zh-Hant.md
└── .auto-readme-work/
    ├── 20260228_184104/
    ├── 20260301_064213/
    ├── 20260301_064740/
    ├── 20260301_065835/
    ├── 20260301_070633/
    ├── 20260302_120620/
    ├── 20260302_124338/
    ├── 20260302_140150/
    └── 20260302_140358/

At this stage, architecture means documentation pipeline architecture, not runtime service architecture.

flowchart LR
    A[Capture run context\n.auto-readme-work/<run-id>/pipeline-context.md] --> B[Analyze repository state\nrepo-structure-analysis.md]
    B --> C[Draft canonical README\nREADME.md]
    C --> D[Align i18n READMEs\ni18n/README.*.md]
    D --> E[Validate quality gates\nlinks, duplicates, structure parity]

• Sear creation tools: applied during content construction to keep sections concrete, complete, and repository-accurate.
• Self-healing tools: applied during validation to remove duplicate blocks, repair stale run references, and restore structural parity.
• Chain of prompt tools: applied across artifacts so each generation stage remains explicit and auditable.

• git
• POSIX shell (examples use bash)
• Markdown-capable editor

• No runnable service or application manifest is present in this repository snapshot.
• Installation, build, and start guidance is therefore documentation-workflow oriented.

No binary package or runtime build step is defined yet.

git clone [email protected]:lachlanchen/AgInTi.git
cd AgInTi

Current usage centers on documentation maintenance and multilingual synchronization.

ls -la
ls -la .auto-readme-work/20260302_140358
ls -la i18n

1. Read .auto-readme-work/20260302_140358/pipeline-context.md.
2. Verify language selector templates in language-nav-root.md and language-nav-i18n.md.
3. Update README.md incrementally as the source of truth.
4. Align i18n/README.*.md files to the same structure and key technical details.
5. Confirm there is exactly one banner and exactly one support panel.

No runtime configuration exists yet. Documentation behavior is driven by repository artifacts.

• pipeline-context.md: run goals and constraints.
• repo-structure-analysis.md: snapshot evidence and gaps.
• language-nav-root.md and language-nav-i18n.md: navigation consistency.
• translation-plan.txt: locale targets and mapping.

cat .auto-readme-work/20260302_140358/language-nav-root.md
cat .auto-readme-work/20260302_140358/language-nav-i18n.md

cat .auto-readme-work/20260302_140358/translation-plan.txt

find . -maxdepth 2 \
  \( -name package.json -o -name pyproject.toml -o -name go.mod -o -name Cargo.toml -o -name pom.xml \)

• Preserve substantive sections and links from canonical README history.
• Prefer incremental edits over destructive rewrites.
• Keep one banner and one support block only.
• Keep root and i18n README structures synchronized.
• Clearly state assumptions whenever runtime or infrastructure details are unknown.
• Apply the philosophy triad as active guardrails:
  • Sear creation tools for high-signal drafting.
  • Self-healing tools for consistency repair.
  • Chain of prompt tools for reproducible handoff between pipeline stages.

That is expected for the current bootstrap phase.

Use the canonical templates in:

• .auto-readme-work/20260302_140358/language-nav-root.md
• .auto-readme-work/20260302_140358/language-nav-i18n.md

git fetch origin
git pull --ff-only

Add build and run instructions only after introducing concrete manifests (for example: package.json, pyproject.toml, go.mod, Cargo.toml) and confirming their paths in this repository.

1. Strengthen sear creation tools with standardized README drafting templates, section quality gates, and clearer evidence-to-output checks.
2. Expand self-healing tools with automated checks for duplicate blocks, locale drift, broken internal anchors, and stale run references.
3. Formalize chain of prompt tools across run stages for reproducible context, generation, translation, and verification traces.
4. Add a single-command documentation maintenance flow once repository scripts are introduced.
5. Add CI checks for markdown quality, link integrity, and i18n structure parity.
6. Introduce concrete runtime components when source manifests and entrypoints are added.
7. Publish a stable licensing decision and add a standalone license file.

Contributions are welcome.

1. Open an issue describing the intended change.
2. Create a focused branch.
3. Keep documentation edits incremental and repository-accurate.
4. Preserve important links, commands, and substantive historical context.
5. Open a pull request with concise verification notes.

git checkout -b docs/your-update
# edit README.md and/or i18n/README.*.md
git add README.md i18n/README.*.md
git commit -m "docs: refine README content"
git push -u origin docs/your-update

This repository includes these root submodules:

• AutoAppDev
• AutoNovelWriter
• OrganoidAgent
• PaperAgent
• LifeReverseEngineering

Nested structure for LRE:

• AgInTi/LifeReverseEngineering/learn -> LazyLearn
• AgInTi/LifeReverseEngineering/earn -> LazyEarn
• AgInTi/LifeReverseEngineering/IDEAS -> IDEAS

AgInTi includes a LAB-compatible prompt-tool bundle under:

• AgInTi/lab_prompt_tools/

Core LRE and AutoAppDev scripts:

• lab_prompt_tools/lre/run_lre_deep_research.sh
• lab_prompt_tools/lre/run_lre_autoappdev.sh
• lab_prompt_tools/lre/prompt_lre_profile_research.sh
• lab_prompt_tools/lre/prompt_lre_book_finder.sh
• lab_prompt_tools/lre/prompt_lre_investment_finder.sh
• lab_prompt_tools/lre/prompt_lre_ideas_finder.sh
• lab_prompt_tools/lre/prompt_lre_self_evolve.sh
• lab_prompt_tools/lre/prompt_lre_repo_autoreadme.sh
• lab_prompt_tools/lre/prompt_lre_repo_autowebsite.sh

Entry point:

• lab_prompt_tools/lre/run_lre_deep_research.sh

Pipeline stages:

1. Profile research stage runs prompt_lre_profile_research.sh.
2. Books stage runs prompt_lre_book_finder.sh using profile output.
3. Investments stage runs prompt_lre_investment_finder.sh using profile output.
4. Ideas stage runs prompt_lre_ideas_finder.sh using profile output.
5. Single-copy report is rebuilt into one markdown file, not appended.
6. Outputs are synced into LRE sub-repos as latest snapshots and self-heal logs.
7. AutoAppDev stage runs for README and website updates across all related repos.
8. Email stage drafts/sends final summary via runtime mail tools.

Key data flow:

• Base output root: ~/.openclaw/workspace/LRE/
• Per-stage JSON output: .../codex/latest-result.json
• Single-copy report: LifeReverseEngineering/notes/lre_single_copy.md
• Learn notes: LifeReverseEngineering/learn/notes/lre_books_latest.md
• Earn notes: LifeReverseEngineering/earn/notes/lre_investments_latest.md
• IDEAS notes: LifeReverseEngineering/IDEAS/notes/lre_research_ideas_latest.md
• Self-heal snapshots: LifeReverseEngineering/**/tools/lre/*_self_heal_latest.{json,log}

Failure handling:

• Each major research step uses timed execution and fallback mode.
• Fallback automatically retries with websearch disabled to prevent full pipeline abort.

Quick run:

cd AgInTi
lab_prompt_tools/lre/run_lre_deep_research.sh --model gpt-5.3-codex --reasoning high

Entry point:

• lab_prompt_tools/lre/run_lre_autoappdev.sh

Targets updated in one run:

• AgInTi
• LifeReverseEngineering
• LifeReverseEngineering/learn (LazyLearn)
• LifeReverseEngineering/earn (LazyEarn)
• LifeReverseEngineering/IDEAS (IDEAS)

For each target repo, the pipeline executes:

1. prompt_lre_repo_autoreadme.sh to update README artifacts.
2. prompt_lre_repo_autowebsite.sh to update website artifacts.
3. Run logs are written under ~/.openclaw/workspace/LRE/autoappdev/logs/<run-id>/.

The course/LinkedIn automation stack is in:

• lab_prompt_tools/career/

Primary scripts:

• lab_prompt_tools/career/start_dec_login_session.sh
• lab_prompt_tools/career/selenium_login_bootstrap.py
• lab_prompt_tools/career/prompt_career_tool_builder.sh
• lab_prompt_tools/career/prompt_career_tool_autodev.sh

Runtime dependencies reused from shared tools:

• lab_prompt_tools/runtime/codex-json-runner.py
• lab_prompt_tools/runtime/codex-noninteractive.sh
• lab_prompt_tools/websearch/prompt_web_search_immersive.sh

Course agent pipeline:

1. Manual bootstrap opens browser session and handles login checkpoints.
2. Builder stage generates a structured tool plan JSON for the objective.
3. Autodev stage consumes the plan and applies/fixes scripts under a target root.
4. Prompt-driven tool generation is iterative, so the same loop can refine broken actions.
5. Output artifacts are stored in run directories under ~/.openclaw/workspace/AgInTi/.

flowchart TD
  A[Human run start_dec_login_session.sh] --> B[selenium_login_bootstrap.py]
  B --> C{Login state}
  C -- logged in --> D[Open DEC course]
  C -- needs auth --> E[Username/password + OTP checkpoint]
  E --> D
  D --> F[Course auto-run loop]
  F --> G[Detect state and question]
  G --> H[Pick answer and click confirm/next]
  H --> I{Progress?}
  I -- yes --> F
  I -- no --> J[Stall handling: retry/refresh/no-action]
  J --> F
  F --> K{Completion marker}
  K -- yes --> L[completed]
  K -- no --> M[incomplete + diagnostics]
  L --> N[sessions/<run-id>/ screenshots + bootstrap.log]
  M --> N

flowchart LR
  O[Human objective] --> P[prompt_career_tool_builder.sh]
  P --> Q[Structured tool plan JSON]
  Q --> R[prompt_career_tool_autodev.sh]
  R --> S[Apply/fix scripts and prompts]
  S --> T[Re-run bootstrap]
  T --> U[New logs/screenshots]
  U --> V[Shadow-self style diagnosis and next repair]
  V --> R

Default target root for generated career tooling:

• AgInTi/AutoAppDev/CareerOps

Donate	PayPal	Stripe

TBD. A standalone license file is planned but is not yet present in the current snapshot.