Exogram + ChatGPT

1. 1The Exogram Chrome extension will be available on the Chrome Web Store shortly.
2. 2Click the Exogram icon and sign in with your account
3. 3Navigate to chatgpt.com and start a conversation
4. 4Right-click any text → "Save to Exogram" to commit facts to your ledger
5. 5Click the Exogram sidebar icon to inject verified context into your prompt

The extension automatically detects ChatGPT's interface and injects a context panel. No API key needed — your browser session handles authentication.

1. 1Open ChatGPT → Explore GPTs → Create a GPT
2. 2In the Configuration tab, add a new Action
3. 3Import the OpenAPI schema from: https://api.exogram.ai/openapi.json
4. 4Set the authentication to API Key (Bearer) and paste your Exogram API key
5. 5In the GPT Instructions, add the system prompt below

You have access to the Exogram semantic ledger. Use it to:
1. STORE facts the user tells you using /v2/vault/store
2. SEARCH for relevant context before answering using /v2/vault/search
3. CHECK for conflicts before accepting new claims
4. EVALUATE proposed actions through /v2/actions/evaluate before executing them

Always search the ledger first. Never make unwarranted inferences — if Exogram has no data, say so.

This method creates a persistent Custom GPT that always has access to your Exogram ledger. Share it with your team for consistent, truth-grounded AI responses.

1. 1Open any ChatGPT conversation
2. 2In your Custom Instructions or system prompt, reference your Exogram data
3. 3Use the Exogram API to fetch context before sending prompts
4. 4Pipe the results into your ChatGPT messages

# Python example: enrich ChatGPT prompts with Exogram context
import requests, openai

# 1. Fetch relevant context from Exogram
exo = requests.post(
    "https://api.exogram.ai/v2/vault/search",
    headers={"Authorization": "Bearer YOUR_EXOGRAM_KEY"},
    json={"query": "user preferences and constraints", "top_k": 5}
).json()

# 2. Build the enriched prompt
context = "\n".join([r["claim"] for r in exo.get("results", [])])
messages = [
    {"role": "system", "content": f"Verified facts:\n{context}"},
    {"role": "user", "content": "Summarize what we know about the project."}
]

# 3. Call ChatGPT with grounded context
response = openai.chat.completions.create(
    model="gpt-4o",
    messages=messages
)
print(response.choices[0].message.content)

This approach works with any OpenAI API integration and gives you full control over how context is injected.