Completion has several dimensions
implementation_complete, tests_sufficient and requirements_satisfied are separate questions. ready_to_finish is not a replacement for directly inspecting test output and requirements.
Inspect the supporting source ↗REPOSITORY ANALYSIS
foreman
An independent supervisor loop that reads worker diffs, logs and tests, asks Jev Nouls about stuck / off-track / verify, then applies a Python policy.
E2 · source reviewedPublic material or source evidence was reviewed. A fixed-version source review requires checking a pinned revision; this is not a reproduction, performance endorsement, or safety guarantee.
- Stars
- 540
- Forks
- 38
- License
- MIT
- Last commit
- 2026-09-22
What Jev does here
AsyncTypeSafeClient.system_one with default jev-latest sends Noul questions for supervision.
Supervision does not write the worker's code. Distinct from cataloged Shifty-Eye-Games/foreman-jev.
Pinned public source was reviewed. This site did not execute the project or independently measure performance or safety.
What task does this project fit?
Supervise coding workers using structured assessments while keeping interventions in deterministic code.
Material analysed：
From input to action
- A bounded snapshot includes job context, diffs, output tails, tests and prior events.
- One async Jev request evaluates nine Noul dimensions such as progress, stuck state and need for verification.
- Python policy translates assessments into continue, worker, verifier or escalation actions.
Implementation details that matter
Probability parsing is part of the contract
jev.py validates all nine values and clamps minor numeric overshoot. Missing or malformed assessment data must follow the failure path rather than appear as successful supervision.
Inspect the supporting source ↗When to choose it
Suitable as a supervisor around a worker harness. It is not an autonomous replacement for tests, human authorization or the worker’s implementation model.
How to inspect it yourself
Read the assessment list in README and normalize_assessment/parse_jev_response in jev.py. Compare expected keys with returned keys; inspect what happens when a required probability is absent.
Limits of this review
- No worker or Codex session was started.
- This is thruwire/foreman, distinct from similarly named Foreman integrations.