Summary
Adds a Python example showing how to govern Composio tool execution with SidClaw — an open-source approval and audit layer for AI agents.
The example uses Composio's native before_execute / after_execute modifier API, so it integrates directly with composio.tools.execute() without any changes to Composio internals.
What it adds
python/examples/sidclaw_governance.py — a standalone script showing:
- Setting up
create_composio_governance_modifiers() from SidClaw
- Passing the modifier functions to
composio.tools.execute()
- SidClaw's automatic slug mapping:
GITHUB_CREATE_ISSUE → operation: "create_issue", target_integration: "github"
- Handling
ActionDeniedError when policy denies or requires approval
Governance outcomes
| Policy decision | Behavior |
|---|
allow | Tool executes immediately. Outcome recorded to hash-chain audit trail. |
approval_required | Execution pauses until a human approves in the SidClaw dashboard. |
deny | ActionDeniedError raised. Tool never executes. |
Why this is useful for Composio users
Composio gives agents access to 1000+ tools. As agents move to production — especially in regulated industries (financial services, healthcare) — teams need a governance layer that decides whether an agent should actually be allowed to execute a given Composio action, and can produce an audit trail of every execution.
SidClaw fills that gap using Composio's own modifier API, so it's fully compatible with any existing Composio setup.
Links