Strategies Library — task upgraded to an ongoing project

Task: SMART-DEBT-Strategies-Library (SDC/IP). 9 rounds, 2026-09-20 → 2026-09-23. The full transcript is in git history (git log -p -- SDC/IP/Tasks/SMART-DEBT-Strategies-Library.md). Outcome: the task’s goal was met. On 2026-09-23 the CEO decided the work is ongoing, not finite, so it was upgraded to the ongoing project Strategies-Library instead of being closed with a successor task.

Goal: “Create SSOT for ALL $MART DEBT strategies”

Task section	Result
1. Where	Done. SDC/IP created as a top-level dept on the CEO’s instruction. The library lives at SDC/IP/Strategies/. The later move to monorepo/sites/sdc for public sharing is on the project ROADMAP.
2. Schema	Done, v0.2 → v0.5. Frontmatter schema with a signed risk-level scale (-1 to 3, 3 = High risk), code, rank/group-rank, and complexity, which feeds the future effort rubric. Deliberately simplified on the CEO’s call. Vocabularies are parsed out of SCHEMA.md §8, so there is a single place to edit them. The library is a conformant Open Knowledge Format (OKF) bundle.
3. Add strategies	32 strategies: 25 in the original build, plus 7 from the Market Drop Wins (MDW) fold-in session 1. Each has concise semantic names, and every claim traces to a named source or reads unverified — needs Talbot. The rest of MDW (not-debt, Harmful / Not Worth It, practice-tier) continues as sdc-ip-mdw-strategy-foldin.
4. Viewing	Done. library-view.html is generated by strategy-lint artifact and is read-only. It has a Risk ladder (lowest risk first) and a Worth It view (a 4×4 effort × benefit matrix), plus an audience filter, a name/code toggle, and the brand entity with the Inter font. Worth It positions are placeholder-derived and bannered until effort/benefit are set.

Key decisions (CEO unless noted)

• IP is a department. The strategy library is the product, not a strategic plan, so it moved from SDC/Strategy/Library. The Risks (Risk-Challenger) gate moved with it and is restated in the charter.
• Risk axis: tier: R0–R4 → signed risk-level. Level 4 was collapsed into 3. The §2.4 Rule 1 captions are the CEO’s own wording, which separates considering from acting; Risks has not yet reviewed them.
• Simplification: leverage was dropped, overruling the defence, and three description axes were cut with complexity kept. availability/decline-type were restored in v0.5 as optional fields for the MDW batch.
• The leverage drop opened a real gap. Two level-1 borrowing strategies escape the risk-level >= 2 Risks filter. The proposed fix is an added criterion, “OR the mechanism involves borrowing”, which goes to Risks first and then the CEO.
• Viewer: audience is an independent filter, not a view. The scatter plot was rebuilt as a matrix. Generated files are left chmod 444, which Obsidian respects; this was generalized into GlobalDevRules.md §3.13.
• Deferred decisions live in Talbot-Later.md, a table rather than task files, reviewed monthly via Routines.md and linked from SDC/DASHBOARD.md.
• 2026-09-23, name: always “$MART DEBT Strategies Library”, plural. Borrowing to invest is dozens of strategies, not one, and consumer debt has dozens more. Recorded in SDC/IP/JOB_DESCRIPTION.md → Scope 1.
• 2026-09-23, construct: the task became an ongoing project.

The 2026-09-23 conversion — what changed

• Rename. strategy-lint’s generator was changed (strategy_lint.py, artifact.py) and all 11 indexes plus the viewer regenerated, so the diff is rename-only. The rename was then swept through the living docs: IP charter, SVP-IP, SDC/DASHBOARD, Strategic-Plan, Strategy ROADMAP, JOB_DESCRIPTION, Filter+Focus, ICP, Better-Rates-Strategy, the strategy-lint util note, and the fold-in task title.
  • Deliberately left singular: dated logs, the ChatGPT brainstorming transcript (verbatim), MDW’s own “Strategy Library v2” (a different artifact’s title), generic compounds like “debt-strategy library”, the quoted CEO sentence in the charter, and task filenames such as sdc-strategy-library-v01 (renaming them breaks links).
• Project stood up at SDC/IP/Projects/Strategies-Library/, with DASHBOARD, STATUS, ROADMAP, UPGRADES and LESSONS, kept outside the OKF bundle so it stays conformant.
• Talbot-Later.md folded into ROADMAP (on Talbot’s suggestion, same day). Its open rows became ROADMAP’s ## Waiting on Talbot section, and its answered rows moved into this log. One file now answers both what is next and what is waiting on me.
• Moved in with git mv: sdc-ip-mdw-strategy-foldin.md (tagged project: Strategies-Library, project-type: ongoing). The charter’s UPGRADES were migrated to the project UPGRADES.md, and the charter now points there.
• Every view repointed: Core DASHBOARD Rock #1, SDC/DASHBOARD, the _active.md tracker, the Ongoing-Projects registry, the Smart task template (kbProjects), Routines.md, SCHEMA §2.4, the charter, and sdc-strategy-library-v01’s Now line.

CEO decisions answered (moved from Talbot-Later.md, 2026-09-23)

Answered	Decision	Outcome
2026-09-21	Folder name: Strategies or Strats	Strategies — “external semantic clarity better”. No rename; code (IL-1, DS-2) is the internal short form.
2026-09-21	Cut the four description axes	Cut all but complexity. capital-required, liquidity-impact and reversibility removed in schema v0.4. complexity stays as the input the effort rubric derives from.
2026-09-21	Keep or drop the leverage flag	Dropped in v0.4, overruling the defence. The Risks gate now keys on risk-level >= 2. ⚠ The residual gap this created is no longer hypothetical — §2.3’s test came back positive on 2026-09-22 and the gap opened at level 1. Whether the field comes back is the first Open row above; it is now a live decision, not a recorded caveat.
2026-09-21	Restore availability and decline-type?	Restored in schema v0.5 (2026-09-22) as optional fields — an omitted availability means always, so the 25 records written before v0.5 stayed valid untouched. The Market Drop Wins fold-in is the batch that made them load-bearing, exactly as §7 predicted. decline-type now renders in the viewer as a “fails in B” warning tag, so the caveat reaches the reader rather than only the file.

Commits

d62df8e (dept + schema v0.2) · 158f643 (v0.3 + OKF + moves) · 7502f1a (v0.4) · df8f55f (subfolders, read-only, viewer) · 89a99fe (viewer round 2) · 8768286 (final viewer + §2.4) · d96f0fb/d0f0787 (MDW fold-in session 1, done under the spun-out task) · plus the 2026-09-23 conversion commit.

LESSONS (re-weighed for context cost; one approved and promoted to the project LESSONS.md)

Six candidates were proposed. The CEO asked for them to be re-weighed, because every line in a LESSONS file is context that loads every session. One survives:

• Before cutting a field as redundant, check two things: whether a lint rule forced the correlation, and whether an external gate keys on the field. leverage looked 25/25 collinear with risk-level only because a v0.2 lint rule forced it. Dropping it opened a compliance gap two strategies wide.

Dropped: the YAML yes/no gotcha is already documented where it’s used (strategy-lint.md, SCHEMA.md). The devtools font stack, matrix-vs-scatter, display:flex vs [hidden], and view-vs-filter lessons are all generic or one-off, and this log already records them.

Round 9 (2026-09-23)

• SMTM skills now find tasks in every business (ai-config dcb3613). The task lookups named Core and MBR by hand, so no SDC task could be found by name. They now use wildcards that cover every business, and audit-skills and audit-blocked were Core/MBR-only and are fixed the same way.
• Ongoing-project continuation assessed. The task-to-task handoff works through the Continuation Gate, but there’s no project-level entry point: no active-task pointer, “terminal” can silently stall a project, and /project-continue handles finite projects only. Talbot approved the design: ROADMAP ## Now plus ## Waiting on Talbot, /task-complete updating ROADMAP, and /project-continue gaining an ongoing branch. Building it is spun out to the KB-OS task below, and the pilot is live in this project’s ROADMAP.
• Memory note left in place: classifier-block-retry-after-explicit-authorization. The vault already records that the classifier can refuse outright, in Core/AI/Claude-Code/Claude-Code.md. What’s only in the memory is the detail that a retry can succeed once explicit written authorization is in context.

Continuation

Successor task: KB-OS-ongoing-project-continuation (Core/Processes/Projects/KB-OS/Tasks/), which builds the approved continuation design. Ongoing project: Strategies-Library. Its ROADMAP ## Now names the active task: /task-continue D:\FSS\KB\SDC\IP\Projects\Strategies-Library\Tasks\sdc-ip-mdw-strategy-foldin.md (5 Next Steps waiting on Talbot, then §4).