docs: spektorAI canonical front door + WhisperForge handbook rename

.company-os/project.yaml

@@ -1,6 +1,6 @@
 project:
   id: "spektorai"
-  name: "WhisperForge"
+  name: "spektorAI"
   type: "repo"
   owner: "kk"
   production_surface: "local-first Streamlit voice-to-knowledge workbench; hosted service mode is optional"

.github/workflows/docs-truth.yml

@@ -4,7 +4,8 @@ on:
   pull_request:
     paths:
       - "Makefile"
-      - "readme.md"
+      - "README.md"
+      - "WHISPERFORGE.md"
       - "STATUS.md"
       - "ROADMAP.md"
       - "changelog.md"

AGENTS.md

@@ -0,0 +1,17 @@
+# AGENTS.md
+
+How to work in this repo, for humans and their agents.
+
+## what this is
+spektorAI is the home of WhisperForge, a local-first Streamlit voice-to-knowledge workbench: audio or dictation in, source-grounded draft + social + Notion page out.
+
+## ground rules
+- match the existing style. surgical changes only.
+- public copy follows the house voice: see kk-brand/VOICE.md. full punk, no em dashes.
+- business logic lives in `whisperforge_core/`. UI and CSS live in `ui/` and `styles.py`. `app.py` is just the composition root.
+- the Makefile is the command surface. `make app` runs it, `make test` runs the suite, `make smoke` boots the shell and hits health, `make docs-check` checks doc truth.
+- routing stays dry-run by default. handoffs to GitHub/Linear/Notion, paid-provider swaps, deploys, and external sends need explicit human approval. do not flip those on your own.
+- no open-source license is granted. treat the code as source-available for collaboration, not freely redistributable.
+
+## the maker
+Kris Krüg (@WalksWithASwagger) · https://kriskrug.co · BC + AI

README.md

@@ -0,0 +1,48 @@
+```text
+K R Ü G  ··  spektorAI
+the future belongs to the weird
+```
+
+# spektorAI
+
+`talk your thoughts. get back a draft in your voice, with receipts.`
+
+spektorAI is the home of WhisperForge, the voice-to-knowledge workbench I use to turn spoken thoughts into publishable work without losing my own voice. I dictate or upload audio, it transcribes, pulls the wisdom, drafts an article, writes the social posts and image prompts, and ships the whole bundle to a structured Notion page. it runs local-first as a Streamlit app on my machine. the voice comes from a per-user prompt and knowledge base, so the output sounds like me and not like a content mill.
+
+it is built for people who think out loud and hate doing the same cleanup-and-formatting grind twice. it is not a hosted SaaS. you run it from a checkout, your keys stay where you put them, and nothing leaves the machine unless you send it somewhere on purpose.
+
+## what it does
+
+- take a Wispr Flow dictation, a pasted note, an upload, or a live browser recording, and turn it into one durable capture
+- transcribe locally on Apple Silicon, on CPU, or in the cloud · your call, your tradeoff
+- run the pipeline: clean the transcript, pull chapters and wisdom, draft an outline, write the article, spin up social posts and image prompts
+- draft, critique, revise. a fact-check pass flags claims the transcript does not actually back up
+- write in your voice via prompts + a knowledge base under `prompts/<user>/`, injected on every call
+- review a draft next to its source receipts before anything ships. advisory scorecards, no gatekeeping
+- hand off a capture or output as a GitHub or Linear issue draft. routing stays dry-run until you approve it
+- ship a clean, color-coded Notion page with title, tags, summary, and run metrics
+- pick your writing engine: OpenAI, Anthropic, or local Ollama. default is Claude Haiku 4.5, fast and cheap and in your voice
+
+## start here
+
+needs Python 3.10+ and `ffmpeg` on your PATH (`brew install ffmpeg` on macOS).
+
+```bash
+git clone https://github.com/WalksWithASwagger/spektorAI.git
+cd spektorAI
+python3 -m venv venv && source venv/bin/activate
+pip install -r requirements.txt
+cp .env.example .env   # add OPENAI_API_KEY + NOTION keys, ANTHROPIC_API_KEY recommended
+make app               # → http://localhost:8501
+```
+
+full setup, the provider matrix, and the run-recovery flow live in the engine handbook, [`WHISPERFORGE.md`](WHISPERFORGE.md). roadmap: `ROADMAP.md`. current handoff state: `STATUS.md`.
+
+## these repos run on agents
+
+built to be operated by humans and machines. see `AGENTS.md` and `llms.txt`.
+
+---
+> i don't build systems that optimize humans. i craft spaces where humans can be gloriously inefficient, creative, and alive.
+
+made by Kris Krüg · [@WalksWithASwagger](https://github.com/WalksWithASwagger) · [kriskrug.co](https://kriskrug.co)

llms.txt

@@ -0,0 +1,20 @@
+# spektorAI
+
+> WhisperForge: a local-first voice-to-knowledge workbench. talk or upload audio, get back a transcript, extracted wisdom, an article, social posts, and image prompts in your own voice, shipped to a structured Notion page.
+
+spektorAI is the canonical home of WhisperForge, a Streamlit app that runs on your own machine. Wispr Flow dictations, pasted notes, uploads, and browser recordings become durable captures, run through a draft-critique-revise pipeline with a source-grounded fact-check, and export to Notion, Markdown, or a GitHub/Linear handoff draft. voice comes from per-user prompts + a knowledge base. Python. built by Kris Krüg / BC + AI.
+
+License posture: no open-source license is currently granted. source-available for project collaboration only. do not treat the archived WalksWithASwagger/whisperforge MIT license as inherited.
+
+## Key files
+- [README](README.md): what spektorAI is, what WhisperForge does, and the one command that proves it
+- [AGENTS.md](AGENTS.md): how agents should operate in this repo
+- [WHISPERFORGE.md](WHISPERFORGE.md): detailed WhisperForge engine docs (features, architecture, setup, env reference)
+- [ROADMAP.md](ROADMAP.md): stabilization and product direction
+- [STATUS.md](STATUS.md): current handoff state
+- [changelog.md](changelog.md): release history
+
+## Links
+- Author: Kris Krüg · https://kriskrug.co
+- Community: BC + AI · https://bc-ai.ca
+- Repo: https://github.com/WalksWithASwagger/spektorAI

scripts/docs_truth_check.py

@@ -10,7 +10,8 @@
 
 
 PROJECT_DOCS = (
-    "readme.md",
+    "README.md",
+    "WHISPERFORGE.md",
     "STATUS.md",
     "ROADMAP.md",
     "changelog.md",