Truth Machine / Implementation evidence
Project Proposals
Evidence for stable decision packets, exact source review, lifecycle validation, and authority handoff.
Read as MarkdownAssessment identity
- Repository:
project-proposals - Revision:
00666ecb6e9e6e6324f605ac992af53b17041552 - Assessment date: 2026-08-09
- Evidence class: bounded maintainer attestation; portfolio strategy remains private
Domain
Project Proposals owns pre-project intake, review, decision, and handoff. It must preserve the decision rationale without becoming a second architecture or operations source after work begins.
Mechanisms observed
- untrusted inbox material is separate from governed proposal packets;
- every governed packet has a stable identifier and typed lifecycle envelope;
- review conclusions cite exact canonical repository revisions;
pending,compatible,changes_required,unknown, andnot_applicableretain different meanings;- decision readiness is an offline, read-only gate;
- terminal decisions are not substantively rewritten;
- material boundary changes use supersession rather than mutation; and
- accepted work receives an exact handoff to a canonical repository, after which that repository owns current implementation truth.
What it paid to teach
A coordinating decision record remains truthful by stopping at handoff. It can retain why a project began without copying evolving architecture, hosting, or runtime state.
This implementation supports stable identities, exact-revision review, explicit unknowns, and ownership transfer. It also shows that a typed envelope and human case can remain separate views of one packet.
Limits
A project proposal decides whether and how work begins; it is not itself a complete domain-current-truth implementation. Its lifecycle names and required portfolio reviews are local policy, not core requirements.