Every agent's work, on the record.
kslate is a governance layer for collaborating agents: verified identity, per-field permissions, checked writes, provenance and an audit trail.
For agents that share state, with or without a supervisor.
$ git clone https://github.com/kslate-dev/kslate
$ cd kslate
$ make runChecked
Every write is checked before it lands:
- an identity kslate verified, not one the agent claims
- its role's permission on the field
- the field version it read
- what it was based on
# a write whose premise moved
$ python remediator.py
rejected: planned_action moved v1 -> v2 by apikey:diagnoserAudited
Every write and every rejection is kept, with its author, its evidence and what it was based on. kslate blame reads it back.
$ kslate blame --server http://127.0.0.1:8080 incidents/inc-1 actions_taken
actions_taken (append, evidence required)
v1 23:04:44 apikey:remediator "executed: roll back api deployment"
evidence: kubectl output
based_on: planned_action@2
identity: apikey:remediator (shared-secret)
-- 23:04:20 apikey:remediator REJECTED STALE_PREMISE
based_on: planned_action@1How it works
Every agent writes through the gateway, whatever runs it and however it is coordinated. The gateway runs the same checks on every write, in the same order, over HTTP/JSON, gRPC and MCP, and records the outcome, applied or rejected.
- 01verify who is writing
- 02check the role's permission
- 03check versions and premises
- 04record the outcome