Manifest
Machine-readable description of the agent capability.
/.well-known/ai-plugin.jsonAgent integration
The public integration surface lets agents use Bradley's AI governance signal-to-POV method inside broader workflows: discover profiles, run the source engine, preview briefs, and preserve the rule that the human keeps final judgment.
Capability
Content Intelligence OS turns AI governance market signals into profile-specific POV candidates, buyer questions, rejected obvious angles, stronger category arguments, and brief previews. The value is the method your agent does not need to reinvent: scan the right inputs, score them against the right wedge, critique the first answer, and keep the founder close to belief.
Integration surface
These endpoints are safe to inspect and use for capability discovery, POV generation, and brief preview. The public surface shows how the capability can be integrated without exposing protected send, cron, lead, or feedback routes.
Machine-readable description of the agent capability.
/.well-known/ai-plugin.jsonRead-only API contract for external agents.
/openapi.jsonInstallable Codex-style skill bundle with instructions, references, and helper script.
/skills/content-intelligence-os-skill.zipPublic package metadata for agents and installers.
/skills/content-intelligence-os-skill.jsonSHA256 package hash for integrity checks before installation.
/skills/content-intelligence-os-skill.sha256Stdio MCP server exposing the capability as named read-only tools.
/mcp/content-intelligence-os-mcp.zipMachine-readable MCP package metadata and runtime instructions.
/mcp/content-intelligence-os-mcp.jsonAvailable customer intelligence profiles and wedges.
/api/source-engine/profilesProfile-aware source run with scored POV candidates.
/api/source-engine/run?profileId=agent-control-plane&limit=8Email-ready preview brief with judgment prompts.
/api/source-engine/brief?profileId=agent-control-plane&limit=8Usage principles
The system scans sources, ranks signals, pressure-tests angles, and prepares judgment prompts.
The founder or operator still decides what they believe, what proof is enough, and what gets published.
Always pass a profile ID. Different AI governance companies should get different angles from the same market signal.
Do not call lead capture, feedback, email sending, or cron routes unless Bradley has explicitly authorized it.
Example call
GET https://bradleyfernandes.com/api/source-engine/run?profileId=model-risk-evidence&limit=8Use the returned candidates to help a user review signals, compare angles, inspect pressure tests, and decide what they actually believe before drafting or publishing.
Installable package
Agents or operators can download the bundle, verify the checksum, and extract it into a skill folder named content-intelligence-os.
curl -L https://bradleyfernandes.com/skills/content-intelligence-os-skill.zip -o content-intelligence-os-skill.zip
curl -L https://bradleyfernandes.com/skills/content-intelligence-os-skill.sha256 -o content-intelligence-os-skill.sha256
mkdir -p ~/.codex/skills/content-intelligence-os
unzip -o content-intelligence-os-skill.zip -d ~/.codex/skills/content-intelligence-osThe package includes a SKILL.md workflow, an API reference, OpenAI-facing metadata, and a helper script for fetching profiles, profile-aware runs, and brief previews.
MCP server
The MCP wrapper gives compatible agents four tools: list_profiles, get_source_engine_snapshot, run_pov_engine, and preview_brief. It calls the public read-only API and does not expose protected email, cron, lead, or feedback routes.
curl -L https://bradleyfernandes.com/mcp/content-intelligence-os-mcp.zip -o content-intelligence-os-mcp.zip
mkdir -p ./content-intelligence-os-mcp
unzip -o content-intelligence-os-mcp.zip -d ./content-intelligence-os-mcp
node ./content-intelligence-os-mcp/server.mjsMCP client config:
{
"mcpServers": {
"content-intelligence-os": {
"command": "node",
"args": ["/absolute/path/to/content-intelligence-os-mcp/server.mjs"]
}
}
}