|
| 1 | +# TR-010 — Truth: Planning Index Consistency Review |
| 2 | + |
| 3 | +## Status |
| 4 | + |
| 5 | +Landed |
| 6 | + |
| 7 | +## Linked Legend |
| 8 | + |
| 9 | +- [TR — Truth](../legends/TR-truth.md) |
| 10 | + |
| 11 | +## Linked Invariants |
| 12 | + |
| 13 | +- [I-001 — Determinism, Trust, And Explicit Surfaces](../invariants/I-001-determinism-trust-and-explicit-surfaces.md) |
| 14 | + |
| 15 | +## Context |
| 16 | + |
| 17 | +The planning model now spans several linked surfaces: |
| 18 | + |
| 19 | +- the live backlog |
| 20 | +- landed design history |
| 21 | +- archived backlog history |
| 22 | +- legend summaries |
| 23 | + |
| 24 | +Those files can drift even when the underlying work is correct. The failure |
| 25 | +mode is usually small but costly: |
| 26 | + |
| 27 | +- a landed cycle stays listed in the live backlog |
| 28 | +- a legend summary lags the current design surface |
| 29 | +- archive state and design state stop matching |
| 30 | + |
| 31 | +The repo already says indexes are part of the workflow. This cycle makes the |
| 32 | +review pass explicit enough to run reliably. |
| 33 | + |
| 34 | +## Human Users, Jobs, And Hills |
| 35 | + |
| 36 | +### Users |
| 37 | + |
| 38 | +- maintainers |
| 39 | +- contributors closing cycles or moving planning artifacts |
| 40 | +- reviewers checking whether planning docs tell the truth |
| 41 | + |
| 42 | +### Jobs |
| 43 | + |
| 44 | +- verify that planning surfaces agree before opening or merging a PR |
| 45 | +- catch backlog, design, archive, and legend drift without manual spelunking |
| 46 | +- keep the planning model trustworthy without adding ceremony for its own sake |
| 47 | + |
| 48 | +### Hill |
| 49 | + |
| 50 | +A maintainer can run one short planning-index review and know that the backlog, |
| 51 | +design, archive, and legend surfaces agree with the real lifecycle state. |
| 52 | + |
| 53 | +## Agent Users, Jobs, And Hills |
| 54 | + |
| 55 | +### Users |
| 56 | + |
| 57 | +- coding agents |
| 58 | +- documentation agents |
| 59 | +- review agents |
| 60 | + |
| 61 | +### Jobs |
| 62 | + |
| 63 | +- check planning-surface alignment explicitly instead of inferring around drift |
| 64 | +- know when a planning-index review is required |
| 65 | +- distinguish live work from landed history and archived intent reliably |
| 66 | + |
| 67 | +### Hill |
| 68 | + |
| 69 | +An agent can treat the planning surfaces as one coherent system because the repo |
| 70 | +defines both the review triggers and the minimum consistency checks. |
| 71 | + |
| 72 | +## Human Playback |
| 73 | + |
| 74 | +- Is it obvious when a planning-index review must run? |
| 75 | +- Does the review stay short enough to use on every cycle-closing PR? |
| 76 | +- Does it say exactly what must match across backlog, design, archive, and |
| 77 | + legends? |
| 78 | + |
| 79 | +## Agent Playback |
| 80 | + |
| 81 | +- Can an agent tell when a branch must review planning-index consistency? |
| 82 | +- Can it enumerate the minimum alignment checks without inventing its own list? |
| 83 | +- Does the doctrine stay pragmatic instead of turning into calendar-driven |
| 84 | + ceremony? |
| 85 | + |
| 86 | +## Explicit Non-Goals |
| 87 | + |
| 88 | +- no attempt to build automated linting for every planning surface in this cycle |
| 89 | +- no requirement for a fixed calendar schedule |
| 90 | +- no broad rewrite of unrelated legacy planning artifacts |
| 91 | + |
| 92 | +## Decisions |
| 93 | + |
| 94 | +### Use The Existing Docs Checklist |
| 95 | + |
| 96 | +The planning-index review should live inside |
| 97 | +[docs/DOCS_CHECKLIST.md](../DOCS_CHECKLIST.md), not as a separate maintainer |
| 98 | +ritual disconnected from the normal docs review pass. |
| 99 | + |
| 100 | +### Define Trigger Conditions In Workflow |
| 101 | + |
| 102 | +[WORKFLOW.md](../../WORKFLOW.md) should say when the planning-index review is |
| 103 | +required: |
| 104 | + |
| 105 | +- when a branch changes backlog, design, archive, or legend indexes |
| 106 | +- when a backlog card moves lifecycle state |
| 107 | +- when a cycle-closing PR is preparing the post-merge truth state |
| 108 | +- when drift is discovered on `main` |
| 109 | + |
| 110 | +### Define The Minimum Alignment Matrix |
| 111 | + |
| 112 | +The review should verify, at minimum: |
| 113 | + |
| 114 | +- live backlog items are still pending, in cycle, or carrying unresolved |
| 115 | + follow-on work |
| 116 | +- landed cycle docs appear in `docs/design/` |
| 117 | +- archived backlog history matches landed or retired cards |
| 118 | +- legend current-cycle summaries agree with the backlog and design surfaces |
| 119 | +- empty-state wording does not invent a new house style accidentally |
| 120 | + |
| 121 | +## Implementation Outline |
| 122 | + |
| 123 | +1. Add this cycle doc. |
| 124 | +2. Extend [docs/DOCS_CHECKLIST.md](../DOCS_CHECKLIST.md) with a planning-index |
| 125 | + review section and the minimum alignment checks. |
| 126 | +3. Update [WORKFLOW.md](../../WORKFLOW.md) and |
| 127 | + [CONTRIBUTING.md](../../CONTRIBUTING.md) so the planning-index review is |
| 128 | + both required and discoverable from tracked doctrine. |
| 129 | +4. Archive the consumed backlog card, update the Truth indexes, and record the |
| 130 | + cycle in [CHANGELOG.md](../../CHANGELOG.md). |
| 131 | + |
| 132 | +## Tests To Write First |
| 133 | + |
| 134 | +No new executable tests. |
| 135 | + |
| 136 | +This is a documentation-truth cycle. Verification is: |
| 137 | + |
| 138 | +- direct cross-check of backlog, design, archive, and legend surfaces after the |
| 139 | + edits |
| 140 | +- formatting validation for touched Markdown files |
| 141 | +- whitespace and diff validation |
| 142 | + |
| 143 | +## Risks And Unknowns |
| 144 | + |
| 145 | +- a manual review can still be skipped if contributors ignore the doctrine |
| 146 | +- the repo may eventually want automated linting for some of these checks |
| 147 | +- legacy historical docs outside the legends model can still drift without |
| 148 | + violating this cycle directly |
| 149 | + |
| 150 | +## Retrospective |
| 151 | + |
| 152 | +This is the right follow-through after the earlier Truth cycles. |
| 153 | + |
| 154 | +The workflow already said indexes mattered. What was missing was a clear, |
| 155 | +repeatable pass for keeping those indexes aligned when real branches land. The |
| 156 | +small drift that showed up on `main` after the previous cycle was exactly the |
| 157 | +kind of failure this review is meant to catch quickly. |
0 commit comments