Skip to content

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EDIKT

Enforced Deterministic Inspection & Knowledge Tracking

EDIKT is a general-purpose, governed AI framework centered on Claude Code that adds governance, deliberation, and memory to complex work — without forcing heavyweight process on simple tasks. It operates on a core thesis: the most critical activity in execution is the systematic transition from Current State to Target State, verified through binary-testable criteria.


Directory Structure

Directory Purpose
edikt/ System definitions: directive (governance engine), deliberation (councils), memory (Beads schema), observability (telemetry), skills (domain modules)
.beads/ Active work tracking: backlog items as JSONL files, templates, compaction archives
governance/ Quality gates and records: standards (DoR, DoD, verifier rubric), decisions (ADRs), certifications
runs/ Per-run working state for governed executions (target state, TSC, binding standard, scope changes)
events/ Append-only JSONL audit logs (directive, council, verification events)
runbooks/ Human-readable operational procedures
commands/ Claude Code command specifications (definitions only)

See the Repo Scaffold Spec for the full canonical directory tree.


Governance Modes

Mode Description
Ungoverned Normal assistant behavior. Memory-aware, minimal structure.
Light Governance Formal artifacts for scope and verification. Phase gates for build and verify.
Full Governance Complete artifact set, council deliberation, full phase gates, certification required.

Escalation is always additive — moving to a higher mode means creating additional artifacts, never reformatting existing work.


Quickstart: Running a Governed Run

  1. Start a governed run with edikt:start — this creates a run directory under runs/ and initializes governance artifacts.
  2. Define the target state — populate target-state.yml with context, outcomes, and constraints.
  3. Extract TSC with edikt:tsc — distill outcomes into binary-testable Target State Criteria.
  4. Derive the Binding Standard with edikt:bind — create the verifier checklist from TSC.
  5. Plan the work with edikt:plan — decompose into a Beads backlog (epics → stories → tasks).
  6. Build — Builder produces artifacts per task acceptance criteria.
  7. Verify with edikt:verify — Verifier independently evaluates work against the Binding Standard.
  8. Certify with edikt:certify — generate the certification packet when all TSC pass.
  9. Learn with edikt:learn — capture lessons, compact completed work, close the run.

Key References

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors