ClearFrame is the open-source AI agent protocol built for full auditability, goal alignment monitoring, and operator control. Every reasoning step logged. Every tool call scored. Every credential encrypted.
A single unsandboxed process reads untrusted content, feeds it into the model, and executes whatever the model produces. There is no audit trail, no goal concept, and no operator visibility.
The same process that fetches a malicious webpage can execute shell commands. Indirect prompt injection is architecturally unavoidable.
API keys sit in ~/.env files readable by every installed plugin. One compromised skill → all credentials exposed.
Tool executions are never logged with tamper-evident records. Post-incident forensics is impossible — you cannot reconstruct what happened.
The runtime has no idea what the agent was supposed to do. Goal drift and out-of-scope actions go completely undetected.
Chain-of-thought is never captured. You cannot inspect why an agent made a decision or replay a session for debugging.
Any GitHub account can publish a plugin with zero review, zero signing, zero verification. One malicious plugin infects every user.
ClearFrame is not a patched version of an existing protocol. It is a ground-up redesign around auditability, isolation, and operator control.
Untrusted content ingestion and tool execution are separate sandboxed processes. Indirect prompt injection is eliminated by design.
Agents declare their goal at session start. Every tool call is scored for alignment. Drift triggers auto-pause. Operators approve ambiguous calls via dashboard.
Every chain-of-thought step is captured as structured, queryable JSON with SHA-256 content hashes. Full session replay available.
Every event is cryptographically chained to the previous. Tampering is detectable via clearframe audit-verify. Full forensic reconstruction possible.
Live REST + WebSocket dashboard. See alignment scores, inspect reasoning traces, approve/block queued calls, and verify audit integrity — all in real time.
A direct feature comparison across the capabilities that matter for production AI agent deployments.
| Capability | ClearFrame | OpenClaw | MCP |
|---|---|---|---|
| Reader/Actor process isolation | ✓ Yes | ✗ No | ✗ No |
| Goal alignment scoring | ✓ Every call | ✗ None | ✗ None |
| Agent goal drift detection | ✓ Auto-pause | ✗ None | ✗ None |
| Operator approval queue | ✓ Built-in | ✗ None | △ Manual |
| Full reasoning trace (RTL) | ✓ Structured JSON | ✗ None | ✗ None |
| Tamper-evident audit log | ✓ HMAC-chained | ✗ None | ✗ None |
| Encrypted credential vault | ✓ AES-256-GCM | ✗ Plaintext .env | △ Varies |
| Context feed hashing | ✓ SHA-256 per chunk | ✗ None | ✗ None |
| Live AgentOps dashboard | ✓ REST + WebSocket | ✗ None | ✗ None |
| Auth required by default | ✓ Yes | ✗ Open by default | △ Optional |
| Localhost-only binding default | ✓ Yes | ✗ All interfaces | △ Varies |
| Open source license | ✓ Apache 2.0 | ✓ MIT | ✓ MIT |
ClearFrame is a standard Python package. Install it, declare your agent's goal, and run.
Requires Python 3.11+. Install from PyPI or directly from the GitHub repo.
# From PyPI (once published)
pip install clearframe
# Or install directly from GitHub right now
pip install git+https://github.com/ibrahimmukherjee-boop/ClearFrame.gitThe CLI scaffolds a project with the correct structure.
clearframe init my-agent
cd my-agentEdit agent.py. Every tool call will be scored against this declared goal.
import asyncio
from clearframe import AgentSession, GoalManifest, ClearFrameConfig
from clearframe.core.manifest import ToolPermission
async def main():
config = ClearFrameConfig()
manifest = GoalManifest(
goal="Search for the latest AI safety papers and summarise them",
permitted_tools=[
ToolPermission(tool_name="web_search", max_calls_per_session=5),
],
allow_file_write=False,
allow_code_execution=False,
)
async with AgentSession(config, manifest) as session:
# Every tool call is scored for alignment before execution
result = await session.call_tool(
"web_search",
query="AI safety papers 2026"
)
print(result)
if __name__ == "__main__":
asyncio.run(main())The AgentOps dashboard starts automatically at localhost:7477.
python agent.py
# Verify audit log integrity at any time
clearframe audit-verify
# View recent audit events
clearframe audit-tail --lines 20