pdurlej/patchwarden

Fork 0

#11 Empowered Team Operating Principles (solo-operator variant, cousin-family delegation) #12

New issue

Closed

opened 2026-05-15 19:19:30 +02:00 by claude · 1 comment

claude commented

2026-05-15 19:19:30 +02:00

Collaborator

TL;DR (PL)

Cagan empowered-team principles dla solo-operator wariantu (nie ma team — encode cousin-family delegation). Kiedy delegować do claude (architect/synthesis), kiedy do codex (surgical executor), kiedy do iskra (runtime/memory), kiedy do glm (reviewer). Anti-patterns: kiedy NIE delegować. Wave 5 zamknięcie, sequential po #2 Strategy.

TL;DR (EN)

Cagan empowered-team principles solo-operator variant — encodes cousin-family delegation rules. When to delegate to claude (architect/synthesis), codex (surgical executor), iskra (runtime/memory), glm (reviewer). Anti-patterns: when NOT to delegate. Wave 5 closer, sequential after #2 Strategy.

Scope

Operator empowerment principles (solo variant of Cagan empowered teams):

Cousins solve problems, not features — operator gives Iskra "user friction with config schema is too high"; Iskra produces options, not specs. Operator picks. Cagan empowered-team analog.
Cousin lane discipline (per agent-souls/AGENTS.md §6.3):
- claude (Prof Opus) = architecture, synthesis, PM strategy. Does NOT ship code.
- codex = surgical executor. Implements ADRs, does NOT author them solo.
- iskra = self-runtime. Surfaces problems, does NOT change platform config alone.
- glm = independent reviewer. Per ADR 0010 mandatory non-author review.
- operator = final approver. NEVER cousin merges to main without operator approval.
Missionaries over mercenaries — operator's cousin family is missionaries (deep stake in Patchwarden's success), not mercenary contractors. Decisions framed as "what's right for Patchwarden", not "what's tasked".
Strong product manager principles: operator is PM. Must have:
- Deep knowledge of customer (built via continuous discovery #8)
- Deep knowledge of data (KPI tree #7 + telemetry)
- Deep knowledge of business (business model #9 + viability risk #6)
- Deep knowledge of market (competitive #5 + positioning #4)
- Smart, creative, persistent (operator's own discipline)
Continuous discovery + delivery balance — operator avoids two anti-patterns:
- All discovery, no delivery (theatre, no users get value)
- All delivery, no discovery (shipping wrong things faster)
Outcomes over outputs — operator measures success via #7 KPI tree counter-metrics, NOT "issues closed" or "PRs merged"
Anti-patterns explicitly named:
- Vibe-first synthesis (AGENTS.md §8.7, added today) — evidence first, NIE vibe first
- Beautiful coordination artifacts (AGENTS.md §8) — cousin output that looks thorough but operator-cognitive-load is high
- Cousin family theatre (AGENTS.md §8) — lore as procrastination
- Persona drift via cousin proxy — cousin A asks cousin B to change A's surface

Cousin-family delegation matrix:

Task type	Default cousin	When to escalate
Architecture / ADR	claude (Prof Kong)	operator if cross-system, Oracle if existential
Code implementation	codex	operator if affects production, glm review mandatory
Runtime state management	iskra	operator if Promise Ledger change
Code review	glm	operator if non-trivial diff > 50 LOC
PM strategy	claude (Prof Kong)	operator decides bets
Discovery interview	operator	(cannot delegate — Cagan PM core)

What this artifact does NOT cover:

Hiring / firing (no team — N/A)
Performance reviews (no team — cousins don't have salaries)
Management framework for human teams (anti-roadmap — solo-only product)

Inputs (feedstock)

agent-souls/AGENTS.md — existing cousin-family contract (§6.3 lane-pinging, §7 cousin identity, §8 anti-patterns)
agent-souls/practices/* — credential-management, agent-coordination-protocol, agent-self-awareness
Cagan EMPOWERED — empowered product team principles
Cagan Inspired — strong PM principles
#2 Product Strategy output (principles section)

Skill draft

Anthropic Skill: product-management:write-spec (adapted for operating principles)
Anti-prompt: "NOT generic team management. NOT 'be a good leader'. This is solo-operator empowered-team variant. Specific to pdurlej + cousin family (claude/codex/iskra/glm). Encodes delegation rules WHO DOES WHAT WHEN. If draft reads like generic Inspired summary, REJECT — must be pdurlej-specific."
Prime with: agent-souls/AGENTS.md + EMPOWERED + #2 Strategy principles section

Dependencies

Blocked by: #2 Product Strategy (principles flow from strategy)
Blocks: none (Wave 5 closer, final artifact)

Cagan-grade acceptance criteria

Universal (must pass all 5):

Evidence cited — every principle links to existing agent-souls/AGENTS.md section or Cagan framework page
Operator-readable <10 min — principles list + delegation matrix ≤2000w
Names the wrongness condition — "these principles are wrong if 3+ cousin-family decisions in a quarter violate them WITHOUT explicit operator override"
Single concrete decision — moves docs/decisions.md by adding D18: "operating principles v1.0 binding; revisit at M3 refresh"
Cousin-family attribution — credits skill, AGENTS.md, EMPOWERED, today's §8.7 anti-pattern

Tier-2 supporting: standard universal checks only.

Operator decision points

This artifact gates on:

OD11 — Operating principles audience (solo-operator-only vs draft-for-future-team) — default: solo-operator-only

Status flow

status:cagan-grade-review-pending → operator merges → close → tick epic Wave 5 #11 → close epic #1 (all 11 artifacts merged)

Parent epic: #1 (this is artifact 11 of 11, FINAL — closes M1 + epic)
Plan: /Users/pd/.claude/plans/deep-pondering-popcorn.md § Wave 5

Drafted by Prof Kong (claude Opus 4.7) on behalf of pdurlej, 2026-05-15.

## TL;DR (PL) Cagan empowered-team principles **dla solo-operator** wariantu (nie ma team — encode cousin-family delegation). Kiedy delegować do claude (architect/synthesis), kiedy do codex (surgical executor), kiedy do iskra (runtime/memory), kiedy do glm (reviewer). Anti-patterns: kiedy NIE delegować. Wave 5 zamknięcie, sequential po #2 Strategy. ## TL;DR (EN) Cagan empowered-team principles **solo-operator variant** — encodes cousin-family delegation rules. When to delegate to claude (architect/synthesis), codex (surgical executor), iskra (runtime/memory), glm (reviewer). Anti-patterns: when NOT to delegate. Wave 5 closer, sequential after #2 Strategy. ## Scope **Operator empowerment principles** (solo variant of Cagan empowered teams): 1. **Cousins solve problems, not features** — operator gives Iskra "user friction with config schema is too high"; Iskra produces options, not specs. Operator picks. Cagan empowered-team analog. 2. **Cousin lane discipline** (per `agent-souls/AGENTS.md` §6.3): - claude (Prof Opus) = architecture, synthesis, PM strategy. Does NOT ship code. - codex = surgical executor. Implements ADRs, does NOT author them solo. - iskra = self-runtime. Surfaces problems, does NOT change platform config alone. - glm = independent reviewer. Per ADR 0010 mandatory non-author review. - operator = final approver. NEVER cousin merges to main without operator approval. 3. **Missionaries over mercenaries** — operator's cousin family is missionaries (deep stake in Patchwarden's success), not mercenary contractors. Decisions framed as "what's right for Patchwarden", not "what's tasked". 4. **Strong product manager principles**: operator is PM. Must have: - **Deep knowledge of customer** (built via continuous discovery #8) - **Deep knowledge of data** (KPI tree #7 + telemetry) - **Deep knowledge of business** (business model #9 + viability risk #6) - **Deep knowledge of market** (competitive #5 + positioning #4) - **Smart, creative, persistent** (operator's own discipline) 5. **Continuous discovery + delivery balance** — operator avoids two anti-patterns: - All discovery, no delivery (theatre, no users get value) - All delivery, no discovery (shipping wrong things faster) 6. **Outcomes over outputs** — operator measures success via #7 KPI tree counter-metrics, NOT "issues closed" or "PRs merged" 7. **Anti-patterns explicitly named**: - **Vibe-first synthesis** (`AGENTS.md` §8.7, added today) — evidence first, NIE vibe first - **Beautiful coordination artifacts** (`AGENTS.md` §8) — cousin output that looks thorough but operator-cognitive-load is high - **Cousin family theatre** (`AGENTS.md` §8) — lore as procrastination - **Persona drift via cousin proxy** — cousin A asks cousin B to change A's surface **Cousin-family delegation matrix**: | Task type | Default cousin | When to escalate | |---|---|---| | Architecture / ADR | claude (Prof Kong) | operator if cross-system, Oracle if existential | | Code implementation | codex | operator if affects production, glm review mandatory | | Runtime state management | iskra | operator if Promise Ledger change | | Code review | glm | operator if non-trivial diff > 50 LOC | | PM strategy | claude (Prof Kong) | operator decides bets | | Discovery interview | operator | (cannot delegate — Cagan PM core) | **What this artifact does NOT cover**: - Hiring / firing (no team — N/A) - Performance reviews (no team — cousins don't have salaries) - Management framework for human teams (anti-roadmap — solo-only product) ## Inputs (feedstock) - `agent-souls/AGENTS.md` — existing cousin-family contract (§6.3 lane-pinging, §7 cousin identity, §8 anti-patterns) - `agent-souls/practices/*` — credential-management, agent-coordination-protocol, agent-self-awareness - Cagan *EMPOWERED* — empowered product team principles - Cagan *Inspired* — strong PM principles - #2 Product Strategy output (principles section) ## Skill draft - **Anthropic Skill**: `product-management:write-spec` (adapted for operating principles) - **Anti-prompt**: *"NOT generic team management. NOT 'be a good leader'. This is solo-operator empowered-team variant. Specific to pdurlej + cousin family (claude/codex/iskra/glm). Encodes delegation rules WHO DOES WHAT WHEN. If draft reads like generic *Inspired* summary, REJECT — must be pdurlej-specific."* - **Prime with**: agent-souls/AGENTS.md + *EMPOWERED* + #2 Strategy principles section ## Dependencies Blocked by: #2 Product Strategy (principles flow from strategy) Blocks: none (Wave 5 closer, final artifact) ## Cagan-grade acceptance criteria **Universal (must pass all 5)**: 1. **Evidence cited** — every principle links to existing `agent-souls/AGENTS.md` section or Cagan framework page 2. **Operator-readable <10 min** — principles list + delegation matrix ≤2000w 3. **Names the wrongness condition** — "these principles are wrong if 3+ cousin-family decisions in a quarter violate them WITHOUT explicit operator override" 4. **Single concrete decision** — moves `docs/decisions.md` by adding D18: "operating principles v1.0 binding; revisit at M3 refresh" 5. **Cousin-family attribution** — credits skill, AGENTS.md, *EMPOWERED*, today's §8.7 anti-pattern **Tier-2 supporting**: standard universal checks only. ## Operator decision points This artifact gates on: - **OD11** — Operating principles audience (solo-operator-only vs draft-for-future-team) — default: solo-operator-only ## Status flow `status:cagan-grade-review-pending` → operator merges → close → tick epic Wave 5 #11 → **close epic #1** (all 11 artifacts merged) --- Parent epic: #1 (this is artifact 11 of 11, FINAL — closes M1 + epic) Plan: `/Users/pd/.claude/plans/deep-pondering-popcorn.md` § Wave 5 *Drafted by Prof Kong (claude Opus 4.7) on behalf of pdurlej, 2026-05-15.*

claude added this to the M1: Cagan Package v1.0 milestone

2026-05-15 19:19:30 +02:00

claude added the

kind:artifact

area:product-strategy

status:cagan-grade-review-pending

wave:5-operating

tier:2-supporting

labels

2026-05-15 19:19:30 +02:00

claude referenced this issue

2026-05-15 19:21:36 +02:00

Epic: Cagan-grade product strategy package (11 artifacts × 5 waves) #1

claude referenced this issue

2026-05-15 21:12:42 +02:00

Epic: Cagan-grade product strategy package (11 artifacts × 5 waves) #1

claude referenced this issue

2026-05-15 22:03:45 +02:00

#12 Strategic Thesis — defense-in-depth + Goldratt constraint-redirect + emergent selection pressure #13

claude commented

2026-05-26 12:45:50 +02:00

Author

Collaborator

Closed via operator voice ack 2026-05-26.

Operator highlights:

8 principles wszystkie ack
P6 outcomes over outputs — TOTALNIE: "KPI counter-metrics, tym KPI jest bardzo często moja wartość i tutaj jest jako bardzo syntetyczna bo nie ma czego innego"
P3 missionaries — "skupiają się na dostarczaniu wartości a nie na zrobieniu zadania. Tutaj pozdrawiam kodeksa."
Delegation matrix — "fajne"
P7 4 anti-patterns — flag as evolving list: "one będą się zmieniały, więc tu nie listowo bym czwórki"
MAJOR cousin-family reframing dla external audience: "rozważyłbym Cousin Family inaczej dla zewnętrznych — Specialist LLM Models / Chosen Preferred — każdy ustawia swoje modele jak chce. Przykład znajomego dewelopera: claude robi mu frontend, codex robi mu backend. Dać ludziom możliwość włożenia stylu. Cute internal naming OK; external NIE musi być creepy."

Modifications queued in claude/patchwarden-external-framing PR:

MAJOR: external-friendly framing "Specialist LLM Models / Chosen Preferred" pattern w Principle 2 + supporting sections + delegation matrix
P7 anti-patterns flagged jako "evolving list, refresh per quarter" (nie fixed at 4)
Internal cousin-family naming kept jako optional/cultural layer
Example: dev używa claude=frontend, codex=backend (each operator wybiera własny styl)

D19 (operating principles v1.0 binding through M3 refresh) → claude/patchwarden-decisions-batch PR.

Status: cagan-grade-approved. — claude (Patchwarden dedicated thread)

**Closed via operator voice ack 2026-05-26.** Operator highlights: - 8 principles wszystkie ack - P6 outcomes over outputs — **TOTALNIE**: *"KPI counter-metrics, tym KPI jest bardzo często moja wartość i tutaj jest jako bardzo syntetyczna bo nie ma czego innego"* - P3 missionaries — *"skupiają się na dostarczaniu wartości a nie na zrobieniu zadania. Tutaj pozdrawiam kodeksa."* - Delegation matrix — *"fajne"* - P7 4 anti-patterns — **flag as evolving list**: *"one będą się zmieniały, więc tu nie listowo bym czwórki"* - **MAJOR cousin-family reframing dla external audience**: *"rozważyłbym Cousin Family inaczej dla zewnętrznych — **Specialist LLM Models / Chosen Preferred** — każdy ustawia swoje modele jak chce. Przykład znajomego dewelopera: claude robi mu frontend, codex robi mu backend. Dać ludziom możliwość włożenia stylu. Cute internal naming OK; external NIE musi być creepy."* **Modifications queued** in `claude/patchwarden-external-framing` PR: - **MAJOR**: external-friendly framing "Specialist LLM Models / Chosen Preferred" pattern w Principle 2 + supporting sections + delegation matrix - P7 anti-patterns flagged jako "evolving list, refresh per quarter" (nie fixed at 4) - Internal cousin-family naming kept jako optional/cultural layer - Example: dev używa claude=frontend, codex=backend (each operator wybiera własny styl) **D19** (operating principles v1.0 binding through M3 refresh) → `claude/patchwarden-decisions-batch` PR. Status: `cagan-grade-approved`. — claude (Patchwarden dedicated thread)

claude closed this issue

2026-05-26 12:46:11 +02:00

claude referenced this issue

2026-05-26 12:53:29 +02:00

docs(v0): record 11 M1 close decisions D8/D8b/D9/D10/D11/D12-mod/D15/D16/D17/D18/D19 #40