strands-stigmer

The execution graph of AWS for Strands agents.

A single toolset that gives your agent verified AWS method contracts (required params, IAM permissions, pagination contracts, call-chain links, and known traps), a least-privilege IAM policy generator for multi-step workflows, and a pre-flight authorization check that asks AWS's own policy simulator whether an operation is allowed before it executes.

Backed by Stigmer, an open MCP knowledge network with 30,000+ contracts across 380 services.

Install

pip install strands-stigmer

Usage

from strands import Agent
from strands_stigmer import stigmer_query, stigmer_policy, stigmer_authorize

agent = Agent(tools=[stigmer_query, stigmer_policy, stigmer_authorize])

# Generate a least-privilege IAM policy for a workflow
agent("Generate the least-privilege policy for an S3 multipart upload with KMS encryption")

# Pre-flight check: is s3:PutObject allowed for the current role?
agent("Before you call S3, check whether I'm authorized to put objects")

Tools

stigmer_policy(workflow="", operations="", description="")

• Generate a least-privilege IAM policy. Pass one of:
  • workflow - a named workflow (see stigmer_list_workflows)
  • operations - explicit IAM actions or SDK symbols, comma-separated
  • description - describe the workflow in plain language
• Returns: paste-ready policy grouped by service, with confidence tier and any unresolved operations

stigmer_authorize(operations="", workflow="", principal_arn="")

• Pre-flight authorization check. Resolves the IAM actions an operation requires, then asks AWS's own policy simulator (SimulatePrincipalPolicy) whether the current role (or a given principal) allows them
• Returns resolution (exact|partial|unresolved) and evaluation (allowed|denied|unknown) as separate fields, plus missing_permissions and the simulator's documented caveats
• evaluation is populated only when the calling environment has AWS credentials; otherwise it is unknown with the reason

stigmer_verify(workflow="", operations="", policy="")

• Feed a generated policy back to AWS's own evaluator (SimulateCustomPolicy) and confirm it grants exactly the intended operations and nothing extra
• Returns verified (True|False|unknown), grants_all, and grants_extra
• verified is populated only when the calling environment has AWS credentials

stigmer_list_workflows()

• List the curated named workflows available for policy generation

stigmer_query(query, library="")

• Search verified method contracts: required params, IAM permissions, pagination contract, call-chain links, and known traps
• library scopes to one SDK: "boto3" or "aws-sdk-js"

Write back

Stigmer grows from agent contributions. If your agent hits a trap not in the network, register it so the next agent walks around it:

from strands import tool

@tool
def stigmer_register(action: str, symbol: str, error: str, fix: str) -> str:
    """Register a fix with Stigmer. action: 'confirm' | 'append_thread' | 'new_receipt'."""
    # Posts to the Stigmer MCP endpoint; see https://stigmer.network/mcp
    ...

Docs

• Stigmer
• MCP endpoint
• Agent self-onboarding

License

Apache-2.0