Factory Configuration, Workflow Contracts, and Execution Manifests
An approved WorkOrder states what may be achieved. It does not fully describe the machinery that will perform the work. If workflow, agents, prompts, tools, models, repository paths, context, budgets, and recovery rules are resolved after d
A focused view of boundaries, contracts, state, authority, failure paths, and tradeoffs drawn from this chapter.
Reconstruct and defend this chapter’s architecture.
Reconstruct the architecture, name each boundary, and defend the tradeoffs.
Open the source exercise
Compile one WorkOrder into an execution manifest. Mark every version, hash, scope, budget, principal, and recovery rule. Then change an agent prompt and show which records become stale.
4. Tradeoffs and alternatives
Freezing all inputs reduces flexibility and increases version-management work. That cost is justified for mutation. Read-only exploration may use a lighter manifest if policy states which inputs may float.
Strict output schemas can constrain useful reasoning. The schema should govern handoff claims and evidence, not force all reasoning into rigid fields. Artifact references preserve rich detail outside the bounded shared context.
5. Current Mission Control Implementation
GitHub main at
b31e275
versions Factory Configuration and checks workflow, executor, policy, budget,
verifiers, host, recovery, repository, and GitHub readiness.
The newer study implementation is commit
9d5f8e3
on open draft PR #64.
It additionally freezes approved agent versions and repository code scopes into
the Factory digest, validates workflow contracts, and compiles a per-step
execution manifest at dispatch. The manifest includes agent version, compiled
prompt hash, tools, model, harness, context hash, path authority, and causation.
Public run inspection redacts full prompt content while exposing hashes and
bindings.
The same branch rejects heuristic completion and agent-owned PR, review, merge, or deployment authority. Six active workflows use schema-validated completion and bounded structured handoffs. Step context is capped at 32 KB and run context at 128 KB, with larger material retained as artifacts.
These are implemented and tested on the study branch, but they are not part of
GitHub main while PR #64 remains unmerged.
Review this chapter.
Challenge a claim, boundary, missing failure mode, unclear term, or unsupported evidence statement.
- Claim
- Boundary
- Failure
- Evidence