Skip to content

docs: add contextweaver + agent-kernel reference flow for policy-before-action #92

@dgenio

Description

@dgenio

Context

contextweaver and agent-kernel solve adjacent parts of the same runtime problem:

  • contextweaver narrows what the model sees: tool cards, compiled context, summarized tool results.
  • agent-kernel should enforce what the agent is allowed to do: capabilities, policy, audit, and authorization boundaries.

For ecosystem clarity, the two projects need a concrete reference flow showing that routing/context selection is advisory and policy enforcement still happens before execution.

Goal

Add a docs/reference example showing a tool-using agent flow where:

  1. contextweaver builds a small candidate/tool-card shortlist.
  2. the model or host selects an intended action.
  3. agent-kernel checks the action against capability/policy rules.
  4. only approved actions proceed to execution.
  5. denied or ask-for-confirmation decisions are represented clearly.

Proposed files

  • docs/integrations/contextweaver.md
  • optional examples/contextweaver_policy_flow.py

Acceptance criteria

  • The docs clearly state that contextweaver routing does not grant permission.
  • The example includes at least three outcomes: allow, ask/confirm, deny.
  • The example uses synthetic data only and runs without external services.
  • The flow references Weaver-compatible routing/execution concepts where useful.
  • README links to the integration page from an ecosystem/interoperability section.

Related

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions