jared/cc-os
Watch
1
0
Fork
You've already forked cc-os
 0
Code Issues 378 Pull requests Projects Releases Packages Wiki Activity

decision-system S4: levers and os-me-route in predict #1246

New issue
Closed
opened 2026-10-23 15:17:26 +00:00 by jared · 2 comments
jared commented 2026-10-23 15:17:26 +00:00
Owner
Copy link

Problem

A prediction does not record where its answer can come from or how much a wrong ruling costs. Thus no caller can set the confidence that a row needs. The design names four objects for these labels: AnswerSource, WrongCostAssessor, WrongCost, and RequiredBand (docs/decision-system/levers.md, levers 1 to 3). None of them exists. This ticket is the first of three S4 tickets. S4b adds the evidence-loop objects. S4c adds the parser fields and the os-me-route CLI.

Approach

The model-backed objects use the DecisionModel port that exists today, not the sketch in levers.md. The port has one shape: Client#ask(question, rules) takes one DecisionModel::Question.new(text:, options:) and returns DecisionModel::Answer.new(choice:). The answer has no probability. Thus:

  • AnswerSource#classify(row) asks one question. Its options are %w[rule precedent facts experiment taste], in this order. Its text holds the row text and the five criterion sentences from levers.md, lever 1. It returns AnswerSource::Result.new(source: <choice as a symbol>). Result has no probability field.
  • WrongCostAssessor#assess(row) asks three questions, in this order: hard_to_undo, compounds, stays_hidden. Each question has the options %w[yes no]. Its text holds the row text and the instruction sentence for that key from levers.md, lever 2. The choice yes gives true. The choice no gives false. There is no threshold.
  • An error from the client goes to the caller. No object gives a default label.

The lever tables give these values:

Source max_rung novel?
rule 1 false
precedent 1 false
facts 2 false
experiment 3 false
taste 0 true
Yes answers WrongCost#level RequiredBand#for
0 :low nil
1 :medium 70
2 :high 85
3 :high 85

Row is the input of the model-backed levers. It holds the row text only. O-9 is settled: the levers live in os-me, and os-sdlc calls the os-me-route CLI (S4c). The ADR challenge rejected an ADR.

Required work:

  1. Add Row, a value that holds the row text — check: present Data.define(:text) in plugins/os-me/lib/os_me/row.rb
  2. Add AnswerSource with MAX_RUNG and Result — check: present MAX_RUNG = { rule: 1, precedent: 1, facts: 2, experiment: 3, taste: 0 } in plugins/os-me/lib/os_me/answer_source.rb
  3. Add WrongCost with yes_count and level — check: present LEVELS = { 0 => :low, 1 => :medium, 2 => :high, 3 => :high } in plugins/os-me/lib/os_me/wrong_cost.rb
  4. Add WrongCostAssessor that asks the three yes/no questions — check: present class WrongCostAssessor in plugins/os-me/lib/os_me/wrong_cost_assessor.rb
  5. Add RequiredBand — check: present BY_LEVEL = { low: nil, medium: 70, high: 85 } in plugins/os-me/lib/os_me/required_band.rb

Acceptance criteria

  1. Given a fake adapter that answers rule, precedent, facts, experiment, and taste in turn, when AnswerSource#classify runs once for each answer, then the results have source :rule, :precedent, :facts, :experiment, :taste; max_rung 1, 1, 2, 3, 0; and novel? false, false, false, false, true.
  2. Given the row text Move Gate::StepOutcome to OsSdlc::StepOutcome., when AnswerSource#classify runs, then the adapter gets one question with the options ["rule", "precedent", "facts", "experiment", "taste"], and the question text includes the row text and the sentence Jared ruled the same kind of case in an earlier session.
  3. Given a fake adapter that answers yes, no, yes in turn, when WrongCostAssessor#assess runs, then it returns WrongCost.new(hard_to_undo: true, compounds: false, stays_hidden: true).
  4. Given the row text Delete the OsSdlc::Error root., when WrongCostAssessor#assess runs, then the adapter gets three questions, each with the options ["yes", "no"]. Each question text includes the row text. The first question text includes deletes code or policy. The second includes sets a pattern. The third includes No test, lint cop, or later gate.
  5. Given WrongCost.new(hard_to_undo: false, compounds: false, stays_hidden: false), when level runs, then it returns :low, and yes_count returns 0.
  6. Given each WrongCost with exactly one true field (three cases), when level runs, then each returns :medium.
  7. Given WrongCost.new(hard_to_undo: true, compounds: true, stays_hidden: false) and WrongCost.new(hard_to_undo: true, compounds: true, stays_hidden: true), when level runs, then both return :high.
  8. Given wrong costs at the levels :low, :medium, and :high, when RequiredBand#for runs, then it returns nil, 70, and 85.
  9. Given a fake adapter that answers no, yes, no (the S22 worked example in levers.md), when WrongCostAssessor#assess runs and RequiredBand#for reads the result, then the level is :medium and the band is 70.
  10. Given a fake adapter that raises OsMe::DecisionModel::UnreadableAnswer, when WrongCostAssessor#assess runs, then the error reaches the caller.

Owner sketch

OsMe::Row = Data.define(:text)

class OsMe::AnswerSource
  MAX_RUNG = { rule: 1, precedent: 1, facts: 2, experiment: 3, taste: 0 }.freeze
  CRITERIA = { rule: '...', precedent: '...', facts: '...', experiment: '...', taste: '...' }.freeze
  Result = Data.define(:source) do
    def max_rung = MAX_RUNG.fetch(source)
    def novel? = source == :taste
  end
  def initialize(client); end
  def classify(row); end
end

class OsMe::WrongCost < Data.define(:hard_to_undo, :compounds, :stays_hidden)
  LEVELS = { 0 => :low, 1 => :medium, 2 => :high, 3 => :high }.freeze
  def yes_count; end
  def level; end
end

class OsMe::WrongCostAssessor
  INSTRUCTIONS = { hard_to_undo: '...', compounds: '...', stays_hidden: '...' }.freeze
  def initialize(client); end
  def assess(row); end
end

class OsMe::RequiredBand
  BY_LEVEL = { low: nil, medium: 70, high: 85 }.freeze
  def for(wrong_cost); end
end

Non-goals

  • The evidence-loop objects: ConfidenceBand.claimed, EvidenceLadder, StopRule, StakeReducer, CostModel (S4b).
  • Parser fields, log fields, and the os-me-route CLI (S4c).
  • A probability on AnswerSource::Result. The DecisionModel port gives none.
  • Changes to the DecisionModel port.
  • Routing in os-sdlc flows (S5).

Origin

  • Trigger: User asked for the decision-system build tickets this session. Rewritten 2026-10-23: split #1246 with exact criteria after the #1245 lesson.
  • Improvised this session: none
  • Chain: off-design S3 build ← acceptance criteria too loose ← first ticket draft. Stop at DESIGN (docs/decision-system/levers.md).
  • Root candidate: none found; searched WrongCost, levers, os-me, predict
  • Where: docs/decision-system/build-plan.md S4; docs/decision-system/levers.md levers 1 to 3; plugins/os-me/lib/os_me/
  • Session: 4f3f95ea-3c6f-4d23-a0ea-e349d6ab02a5
  • Transcript: /home/jared/.claude/projects/-home-jared-dev-cc-os/4f3f95ea-3c6f-4d23-a0ea-e349d6ab02a5.jsonl

VERDICT: CREATE (user-approved; backlog search only)

## Problem A prediction does not record where its answer can come from or how much a wrong ruling costs. Thus no caller can set the confidence that a row needs. The design names four objects for these labels: `AnswerSource`, `WrongCostAssessor`, `WrongCost`, and `RequiredBand` (`docs/decision-system/levers.md`, levers 1 to 3). None of them exists. This ticket is the first of three S4 tickets. S4b adds the evidence-loop objects. S4c adds the parser fields and the `os-me-route` CLI. ## Approach The model-backed objects use the `DecisionModel` port that exists today, not the sketch in `levers.md`. The port has one shape: `Client#ask(question, rules)` takes one `DecisionModel::Question.new(text:, options:)` and returns `DecisionModel::Answer.new(choice:)`. The answer has no probability. Thus: - `AnswerSource#classify(row)` asks one question. Its options are `%w[rule precedent facts experiment taste]`, in this order. Its text holds the row text and the five criterion sentences from `levers.md`, lever 1. It returns `AnswerSource::Result.new(source: <choice as a symbol>)`. `Result` has no `probability` field. - `WrongCostAssessor#assess(row)` asks three questions, in this order: `hard_to_undo`, `compounds`, `stays_hidden`. Each question has the options `%w[yes no]`. Its text holds the row text and the instruction sentence for that key from `levers.md`, lever 2. The choice `yes` gives `true`. The choice `no` gives `false`. There is no threshold. - An error from the client goes to the caller. No object gives a default label. The lever tables give these values: | Source | `max_rung` | `novel?` | |---|---|---| | `rule` | 1 | false | | `precedent` | 1 | false | | `facts` | 2 | false | | `experiment` | 3 | false | | `taste` | 0 | true | | Yes answers | `WrongCost#level` | `RequiredBand#for` | |---|---|---| | 0 | `:low` | `nil` | | 1 | `:medium` | 70 | | 2 | `:high` | 85 | | 3 | `:high` | 85 | `Row` is the input of the model-backed levers. It holds the row text only. O-9 is settled: the levers live in os-me, and os-sdlc calls the `os-me-route` CLI (S4c). The ADR challenge rejected an ADR. Required work: 1. Add `Row`, a value that holds the row text — check: present `Data.define(:text)` in `plugins/os-me/lib/os_me/row.rb` 2. Add `AnswerSource` with `MAX_RUNG` and `Result` — check: present `MAX_RUNG = { rule: 1, precedent: 1, facts: 2, experiment: 3, taste: 0 }` in `plugins/os-me/lib/os_me/answer_source.rb` 3. Add `WrongCost` with `yes_count` and `level` — check: present `LEVELS = { 0 => :low, 1 => :medium, 2 => :high, 3 => :high }` in `plugins/os-me/lib/os_me/wrong_cost.rb` 4. Add `WrongCostAssessor` that asks the three yes/no questions — check: present `class WrongCostAssessor` in `plugins/os-me/lib/os_me/wrong_cost_assessor.rb` 5. Add `RequiredBand` — check: present `BY_LEVEL = { low: nil, medium: 70, high: 85 }` in `plugins/os-me/lib/os_me/required_band.rb` ## Acceptance criteria 1. Given a fake adapter that answers `rule`, `precedent`, `facts`, `experiment`, and `taste` in turn, when `AnswerSource#classify` runs once for each answer, then the results have `source` `:rule`, `:precedent`, `:facts`, `:experiment`, `:taste`; `max_rung` 1, 1, 2, 3, 0; and `novel?` false, false, false, false, true. 2. Given the row text `Move Gate::StepOutcome to OsSdlc::StepOutcome.`, when `AnswerSource#classify` runs, then the adapter gets one question with the options `["rule", "precedent", "facts", "experiment", "taste"]`, and the question text includes the row text and the sentence `Jared ruled the same kind of case in an earlier session.` 3. Given a fake adapter that answers `yes`, `no`, `yes` in turn, when `WrongCostAssessor#assess` runs, then it returns `WrongCost.new(hard_to_undo: true, compounds: false, stays_hidden: true)`. 4. Given the row text `Delete the OsSdlc::Error root.`, when `WrongCostAssessor#assess` runs, then the adapter gets three questions, each with the options `["yes", "no"]`. Each question text includes the row text. The first question text includes `deletes code or policy`. The second includes `sets a pattern`. The third includes `No test, lint cop, or later gate`. 5. Given `WrongCost.new(hard_to_undo: false, compounds: false, stays_hidden: false)`, when `level` runs, then it returns `:low`, and `yes_count` returns 0. 6. Given each `WrongCost` with exactly one `true` field (three cases), when `level` runs, then each returns `:medium`. 7. Given `WrongCost.new(hard_to_undo: true, compounds: true, stays_hidden: false)` and `WrongCost.new(hard_to_undo: true, compounds: true, stays_hidden: true)`, when `level` runs, then both return `:high`. 8. Given wrong costs at the levels `:low`, `:medium`, and `:high`, when `RequiredBand#for` runs, then it returns `nil`, 70, and 85. 9. Given a fake adapter that answers `no`, `yes`, `no` (the S22 worked example in `levers.md`), when `WrongCostAssessor#assess` runs and `RequiredBand#for` reads the result, then the level is `:medium` and the band is 70. 10. Given a fake adapter that raises `OsMe::DecisionModel::UnreadableAnswer`, when `WrongCostAssessor#assess` runs, then the error reaches the caller. ## Owner sketch ```ruby OsMe::Row = Data.define(:text) class OsMe::AnswerSource MAX_RUNG = { rule: 1, precedent: 1, facts: 2, experiment: 3, taste: 0 }.freeze CRITERIA = { rule: '...', precedent: '...', facts: '...', experiment: '...', taste: '...' }.freeze Result = Data.define(:source) do def max_rung = MAX_RUNG.fetch(source) def novel? = source == :taste end def initialize(client); end def classify(row); end end class OsMe::WrongCost < Data.define(:hard_to_undo, :compounds, :stays_hidden) LEVELS = { 0 => :low, 1 => :medium, 2 => :high, 3 => :high }.freeze def yes_count; end def level; end end class OsMe::WrongCostAssessor INSTRUCTIONS = { hard_to_undo: '...', compounds: '...', stays_hidden: '...' }.freeze def initialize(client); end def assess(row); end end class OsMe::RequiredBand BY_LEVEL = { low: nil, medium: 70, high: 85 }.freeze def for(wrong_cost); end end ``` ## Non-goals - The evidence-loop objects: `ConfidenceBand.claimed`, `EvidenceLadder`, `StopRule`, `StakeReducer`, `CostModel` (S4b). - Parser fields, log fields, and the `os-me-route` CLI (S4c). - A probability on `AnswerSource::Result`. The `DecisionModel` port gives none. - Changes to the `DecisionModel` port. - Routing in os-sdlc flows (S5). ## Origin - Trigger: User asked for the decision-system build tickets this session. Rewritten 2026-10-23: split #1246 with exact criteria after the #1245 lesson. - Improvised this session: none - Chain: off-design S3 build ← acceptance criteria too loose ← first ticket draft. Stop at `DESIGN` (`docs/decision-system/levers.md`). - Root candidate: none found; searched WrongCost, levers, os-me, predict - Where: docs/decision-system/build-plan.md S4; docs/decision-system/levers.md levers 1 to 3; plugins/os-me/lib/os_me/ - Session: 4f3f95ea-3c6f-4d23-a0ea-e349d6ab02a5 - Transcript: /home/jared/.claude/projects/-home-jared-dev-cc-os/4f3f95ea-3c6f-4d23-a0ea-e349d6ab02a5.jsonl VERDICT: CREATE (user-approved; backlog search only)
jared commented 2026-10-23 17:10:51 +00:00
Author
Owner
Copy link

Work started on branch ticket-1246 (/os-sdlc:implement, S4a).

Work started on branch ticket-1246 (/os-sdlc:implement, S4a).
jared commented 2026-10-23 17:17:51 +00:00
Author
Owner
Copy link

Resolution

Done: S4a built: Row, AnswerSource (+Result), WrongCost, WrongCostAssessor, RequiredBand on the current DecisionModel port; implementation 127 complete.

Evidence: Branch ticket-1246, Runner commit; suite and lint gates green; contract audit pass after one refactor round; 10 acceptance criteria covered in answer_source_test.rb and wrong_cost_test.rb.

Follow-ups: S4b #1262 and S4c #1263 continue S4. A lens proposal on A3 (prefix-sharing assessor under a ticket-pinned name) goes through the lens challenge. S8 #1250 swaps the DecisionModel port to the decision-model gem, which changes AnswerSource and WrongCostAssessor asks.

## Resolution **Done:** S4a built: Row, AnswerSource (+Result), WrongCost, WrongCostAssessor, RequiredBand on the current DecisionModel port; implementation 127 complete. **Evidence:** Branch ticket-1246, Runner commit; suite and lint gates green; contract audit pass after one refactor round; 10 acceptance criteria covered in answer_source_test.rb and wrong_cost_test.rb. **Follow-ups:** S4b #1262 and S4c #1263 continue S4. A lens proposal on A3 (prefix-sharing assessor under a ticket-pinned name) goes through the lens challenge. S8 #1250 swaps the DecisionModel port to the decision-model gem, which changes AnswerSource and WrongCostAssessor asks.
jared closed this issue 2026-10-23 17:17:51 +00:00
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
ticket-1337
No results found.
Labels
Clear labels   
P0
Priority 0 — drop everything

  
P1
Priority 1 — next up

  
P2
Priority 2 — normal

  
P3
Priority 3 — someday

  
bug
Something is broken

  
category/code-change
Ticket category

  
category/cross-repo
Ticket category

  
category/eval-run
Ticket category

  
category/investigation
Ticket category

  
category/no-test-target
Ticket category

  
category/owner-ruling
Ticket category

  
category/prose-artifact
Ticket category

  
category/recurring-chore
Ticket category

  
category/structure

   
category/umbrella
Ticket category

  
crc-refactor
CRC refactor run work record; exempt from the ticket-skeptic

  
create
Ticket creates something new

  
delete
Ticket removes something

  
enhancement
New feature or improvement

  
filed-by/agent

   
filed-by/user

   
frozen
Pre-reset backlog freeze 2026-08-16; do not work unless rediscovered by a live run

  
lint-rule
Ticket proposes/modifies a lint rule or cop

  
needs-info
Waiting on reporter for more information

  
needs-triage
Maintainer needs to evaluate this issue

  
next
Human-curated next up — AI never applies or removes

  
plugin/cc-architect
Scope: the cc-architect plugin

  
plugin/os
Scope: the os plugin

  
plugin/os-adr
Scope: the os-adr plugin

  
plugin/os-aidd-lint
Scope: the os-aidd-lint plugin

  
plugin/os-backlog
Scope: the os-backlog plugin

  
plugin/os-context
Scope: the os-context plugin

  
plugin/os-doc-hygiene
Scope: the os-doc-hygiene plugin

  
plugin/os-sdlc
Scope: the os-sdlc plugin

  
plugin/os-vault
Scope: the os-vault plugin

  
project/cc-os
Parent-scope: cc-os repo-wide, not specific to one plugin

  
ready-for-agent
Triage complete; an agent can pick this up

  
ready-for-human
Human-owned; never picked up autonomously

  
recurring
Recurring flavor: occurrence outcomes commented, never closed by AI

  
review
Shipped semi work awaiting human sign-off

  
rule-proposal
A rule change that survived the gate; Jared reads these when he chooses

  
update
Ticket updates something existing

  
waiting
Blocked; blocker named in a comment

  
wayfinder:grilling

   
wayfinder:map

   
wayfinder:map
wayfinder map ticket

  
wayfinder:research

   
wayfinder:task

   
wayfinder:task
wayfinder map child: task

  
wontfix
Will not be actioned

  
worklist/deviations
Deviation incidents from skill/pipeline runs — evidence pile for correcting the owning artifact

  
worklist/lint-rule
Proposed os-sdlc or os-aidd-lint rule

  
worklist/new-implement-build
Updated implementation harness (ast-grep + stepper)

No labels
P0
P1
P2
P3
bug
category/code-change
category/cross-repo
category/eval-run
category/investigation
category/no-test-target
category/owner-ruling
category/prose-artifact
category/recurring-chore
category/structure
category/umbrella
crc-refactor
create
delete
enhancement
filed-by/agent
filed-by/user
frozen
lint-rule
needs-info
needs-triage
next
plugin/cc-architect
plugin/os
plugin/os-adr
plugin/os-aidd-lint
plugin/os-backlog
plugin/os-context
plugin/os-doc-hygiene
plugin/os-sdlc
plugin/os-vault
project/cc-os
ready-for-agent
ready-for-human
recurring
review
rule-proposal
update
waiting
wayfinder:grilling
wayfinder:map
wayfinder:map
wayfinder:research
wayfinder:task
wayfinder:task
wontfix
worklist/deviations
worklist/lint-rule
worklist/new-implement-build
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
jared/cc-os#1246
No description provided.