git:20260711.55a978c to git:20260711.64106fb

1 added, 0 removed. Audit A to A.

# AGENTS.md

## Scope

These instructions apply to the whole repository.

## Start here

1. Read `.assets/docs/architecture-map.md` to identify the correct layer before editing.

2. Read `.assets/docs/STYLEGUIDE.md` before changing Markdown, examples, templates, or skill references.

3. Read `.skills/architecture.md` before changing runtime skills, provider adapters, export behavior, or install behavior.

4. Load only the relevant platform skill and references for the task.

## Source of truth

- Keep runtime methodology in `.skills/agent-skill/`.

- Keep provider folders thin. Treat `.skills/providers/` as adapter-specific install notes, wrapper commands, and metadata.

- Keep the export and install implementation in `.skills/export/`.

- Keep public project positioning, install commands, and module summaries in `README.md`.

- Do not commit Career Context files, user career exports, screenshots, or unstaged generated install output.

## Brand contract

- Product: **VitaeContext**. Tagline: **Keep your career context. Reuse it across AI.**

- Canonical definition: **VitaeContext gives AI agents a private, reusable source of truth about a person's career, then provides focused skills for turning that context into grounded professional work.**

+ - Position VitaeContext as the context layer before agent-assisted career systems and workflows. It is provider- and system-agnostic: it prepares the precise, evidence-bounded Career Context those systems use, rather than depending on a particular CV builder, application tool, or AI provider.

- Use **Career Context** for the private user-owned artifact. Describe **VitaeGraph** as VitaeContext's optional deeper structured-record layer.

- Frame the product as career-context infrastructure. SEO, AEO, ATS, and platform optimization are capabilities of focused modules, not the umbrella identity.

- Keep `agentkit-seo` only where compatibility behavior or migration history requires it.

- Use the established brand assets under `.assets/image/`. Do not remove historical visual assets solely because they predate the rename.

- Brand colors: `#123C2D`, `#1F684B`, `#91B9A3`, `#E7F0EB`, `#F7F5EF`, `#0D2019`, `#18201C`, `#5C6861`, and `#CED8D2`. Semantic accents: `#287A55`, `#4C7180`, `#B7792D`, `#A14E46`, and `#716B86`.

## Editing rules

- Prefer narrow edits that follow the existing repository structure.

- Keep Markdown plain, factual, and source-aware.

- Separate documented platform behavior from inference.

- Do not invent SEO, ATS, GitHub, LinkedIn, X, or provider ranking guarantees.

- Update dependent docs when package behavior, provider behavior, or public workflows change.

- When writing or updating `wiki/` entries inside `.skills/agent-skill/`, keep them factual, compact, and runtime-oriented. Each wiki file must start with the VitaeContext wiki metadata block, use only the allowed confidence values, update `last_reviewed` when claims are meaningfully checked, and mark stale, disputed, or source-dependent claims instead of presenting them as stable facts.

## Validation

Run the smallest relevant check for the change:

```bash

npm run validate

node .skills/export/scripts/vitaecontext.mjs version

node .skills/export/scripts/vitaecontext.mjs export --provider all --output /tmp/vitaecontext-export --force

npm pack --dry-run

```