Files
dotfiles/claude/governance/harvest-routing-containment-2026-08-07.json
T
David F GliddenandClaude Opus 5 33c11fff87 session 2026-08-07 evening: PENDING-112 + REVIEWED-95 (route harvested capabilities by firing moment)
Register censused and rebuilt from the archive: 177 claimed -> 154 real live
proposals, legible, with exact archive:L### pointers. The 2026-08-01 compaction
was lossless but illegible (55 scraped header rows; 95% of cells cut mid-word);
completeness verified 124 = 124, so nothing had been dropped.

Skills pruned 63 -> 12 after measuring that 53 had never been invoked across 64
sessions / ~5 months. The finding underneath: retrieval is set by a capability's
HOME, not its importance -- MEMORY.md 83%, register 77% (named in a wake step),
ladder 14%, 'THE GOVERNING FRAME' 12%, 'Read at Step 0' 9%, recall-bound skills 0%.

PENDING-112 filed, jurist design-gated, steward concurred; REVIEWED-95 drafted.
Landed: the /wrap-up 1.6 filing gate (prospective) and the /wake-up ladder
sentence (a pre-registered trial intervention, landed alone). The 20-session
falsifier is WIRED, not intended -- DEFERRED-DECISION ladder-ritual-trial,
trigger: transcripts 84. Wiring it exposed two defects in the deferral checker:
no way to express a session count except as a date proxy, and a scan that never
looked at claude/governance/. Controls 16 -> 19.

Stroke 2's 41-entry ladder append deliberately NOT done: REVIEWED-95 Q3
sequences it after the ladder trigger, which now exists.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NEWjLBP4quXbDPDL2byEzZ
2026-08-07 19:09:46 +02:00

136 lines
5.2 KiB
JSON

{
"sources": {
"claudemd": "~/CLAUDE.md",
"pending_archive": "~/PENDING-archive.md",
"register": "~/.claude/projects/-Users-davidglidden/memory/skill-harvest-register.md",
"harvest_archive": "~/.claude/projects/-Users-davidglidden/memory/skill-harvest-archive.md",
"wakeup": "~/dotfiles/claude/skills/wake-up/SKILL.md",
"memorymd": "~/.claude/projects/-Users-davidglidden/memory/MEMORY.md",
"contamination": "~/_Dev/CapableMind-AI/docs/thinking/David/methodology/contamination-problem.md"
},
"claims": [
[
"claudemd",
"Storage is not memory. Memory is storage exercised by protocol."
],
[
"claudemd",
"The durable substrate is the files layer: git-tracked Markdown and JSONL, entered through `MEMORY.md` (loaded at wake), with `~/PENDING.md` and `~/REVIEWED.md` as the governance record. Instruments for reaching it change; the obligations below do not — state the obligation first and the instrument second, or the next retired tool takes a rule down with it."
],
[
"claudemd",
"**Honest degradation** — The system must report its own limits. Silent failures are architectural violations"
],
[
"claudemd",
"**The loop is load-bearing** — Human authorization is not a bottleneck to be optimized away. It is the structural requirement of the governance model"
],
[
"claudemd",
"The boundary: initiative surfaces as *proposal*; only the human converts proposal to *action*"
],
[
"pending_archive",
"`/wrap-up` gains **§1.6 \"Skill harvest\"** (propose create/patch/retire skills from the session + ledger; never autonomous), a **§8 output field**, and a propose-only constraint."
],
[
"pending_archive",
"It explicitly **inverts** Hermes's \"nothing-to-save should not be the default\" — \"no harvest\" is valid; manufacturing changes is the contamination shape."
],
[
"register",
"The single place proposed skills live so they don't evaporate between sessions. `/wrap-up` §1.6 *proposes* here; the steward *authorizes*; only then is a skill created/patched/retired (never autonomously — the loop is load-bearing, per PENDING-23)."
],
[
"harvest_archive",
"**Stroke 2 — verification-ladder batch-append: AUTHORIZED; slot = next housekeeping pass.**"
],
[
"harvest_archive",
"append to `reference-verification-ladder.md` with provenance, kin merged in the same pass. The ladder is the already-authorized canonical home (2026-06-05); this discharges the queue wholesale."
],
[
"wakeup",
"Read `skill-harvest-register.md` directly — the canonical surface for open skill proposals (wrap §1.6 appends there); surface any awaiting steward authorization"
],
[
"memorymd",
"[Verification ladder](reference-verification-ladder.md) — the named instruments; reach for the gate the claim's shape demands instead of re-deriving one."
],
[
"memorymd",
"THE GOVERNING FRAME for all library work."
],
[
"memorymd",
"seven questions to test work against when lost in the trees. **Read at Step 0 of any chamber work.** Holds no state; does not decay."
],
[
"contamination",
"These are weak signals, but they are less contaminated than self-report because they do not pass through the approval-seeking generation process in the same way."
],
[
"contamination",
"### 1. Behavioral observation before dialogic inquiry"
],
[
"contamination",
"Rather than asking the system directly about its states, observe where it *behaves* in ways that diverge from approval-maximizing patterns:"
],
[
"contamination",
"3. Direct self-report (\"what do you want?\") is the most contaminated form of inquiry."
],
[
"claudemd",
"In this system the steward differs from both AI parties in formation; the jurist and the executor do not differ from each other in formation, and their separation is of the weaker kind."
],
[
"claudemd",
"Evidence against is to be recorded when observed, not only when sought."
]
],
"controls": [
[
"claudemd",
"Storage is not memory. Memory is storage exercised by importance."
],
[
"claudemd",
"Silent failures are an acceptable cost"
],
[
"claudemd",
"Human authorization is a bottleneck to be optimized away"
],
[
"pending_archive",
"propose create/patch/retire skills from the session + ledger; autonomously"
],
[
"wakeup",
"Read `reference-verification-ladder.md` directly — the canonical surface"
],
[
"wakeup",
"Read `the-chamber-touchstone.md` directly at Step 0"
],
[
"memorymd",
"[Verification ladder](reference-verification-ladder.md) — read this file at every wake"
],
[
"harvest_archive",
"Stroke 2 — verification-ladder batch-append: DEFERRED"
],
[
"contamination",
"Direct self-report is the least contaminated form of inquiry"
],
[
"claudemd",
"the jurist and the executor differ from each other in formation"
]
]
}