Two files, one repository

Mark393295827/third-brain-v7-skills ships 3 formats across 3 indexed files. The question worth asking is whether the second one says anything the first does not.

CompareAGENTS.md ↔ CLAUDE.md AGENTS.md ↔ GEMINI.md CLAUDE.md ↔ GEMINI.md

A · AGENTS.md · 772 words B · CLAUDE.md · 837 words

What each file covers, counted
Dimension	Shared	Only in A	Only in B	Overlap
Sections	10	5	8	43%
Commands	0	0	0	—
Section tags	4	0	0	100%

What each file covers

Sections

10 shared · 5 only in A · 8 only in B

− Third Brain V7 Skills — Codex CLI
− Skills
− Compatibility
− V7 Promotion Gate
− Adoption Ladder
+ Third Brain V7.1 Skills — Claude Code
+ Skill Categories
+ Usage
+ Recommended Adoption Ladder
+ Engineering Routing Boundary
+ Grounding Principles
+ 3B Creativity Algorithms
+ Agent Teams
📥 Knowledge Pipeline
🔄 Daily Loop
🎨 Behavior & Creativity
🔬 Research & Quality
🔄 Learning
📊 Context & Cost
🏗️ Engineering
💼 Strategy & Operations
Skill Contract
Karpathy LLM OS

Commands

neither file has any

Section tags

4 shared · 0 only in A · 0 only in B

lint-format
api
deployment
agent-behaviour

Line diff

+63 added−48 removed38 unchanged37.6% identical

Mark393295827/third-brain-v7-skills · AGENTS.md

@@ −1 @@

1# Third Brain V7 Skills — Codex CLI

3This repository contains Agent Skills for the Codex CLI environment. Place skills in `~/.agents/skills/`.

5V7 treats Obsidian as durable disk and governance, while every skill uses a profile-aware execution contract. V7.1 adds the 20th skill, `graph-engineering`, for bounded static dependency DAGs. Do not turn wiki signals into rules until they pass the promotion gate: repeated source support or local verification, bounded macro action, preserved provenance/permissions, and a cheap objective check.

7## Skills

9### 📥 Knowledge Pipeline

10- **wiki-ingest** — STOW pipeline with source-risk classification, block refs, clipping lifecycle, understanding gate, promotion controls, and post-ingest lint.

11- **knowledge-ops** — Multi-layer knowledge management with Markdown-first retrieval, optional vector storage, evidence hierarchy, deduplication, Agent/Wiki flywheel, and knowledge debt queues.

12- **wiki-lint** — Wiki health check for graph integrity, provenance, links, clipping lifecycle, understanding, and promotion readiness.

14### 🔄 Daily Loop

15- **daily-okr** — 7 Key Results daily knowledge compound cycle; may consume the scheduled daily knowledge-loop note when present.

16- **cognitive-compile** — 8-section deep learning compile framework.

18### 🎨 Behavior & Creativity

19- **behavior-design** — Behavior change system with HAS framework.

20- **creativity-engine** — Combinatorial ideation + minimum experiments.

22### 🔬 Research & Quality

23- **deep-research** — STOW-compatible research harness with ChatGPT-style preflight, source/claim ledgers, activity trace, citations, privacy checks, and wiki-ingest handoff.

24- **verify-before-claim** — Verification-first quality gate.

26### 🔄 Learning

27- **session-learn** — Knowledge extraction with Closure Protocol.

28- **project-flow-ops** — Project triage and tracking.

30### 📊 Context & Cost

31- **context-manager** — Runtime-derived context budgets, checkpoints, compaction, retrieval, and capability routing.

32- `token-cost-tracker` is a command under `commands/`, not an Agent Skill.

34### 🏗️ Engineering

35- **agentic-engineering** — Agent-as-process workflow refactoring with autonomy defaults, delegated-action boundaries, state checkpoints, write-back, and verification gates.

36- **loop-engineering** — Temporal-depth control through bounded Trigger -> Execute -> Verify -> State loops, durable contracts, hard budgets, and stop/recovery rules.

37- **graph-engineering** — Dependency-width control for bounded static DAGs with explicit dependencies, independent branches, typed joins, and node-local recovery; dynamic and cyclic graphs are out of scope in V7.1.

38- **harness-engineering** — Runtime kernel design: scheduler, permissions, tools as system calls, MCP/Skills/Hooks selection, provenance ledgers, observability, and recovery.

39- **agent-teams-command** — Multi-agent process ownership, IPC, worktree isolation, async budget envelopes, integration, cleanup, and evidence gates.

41### 💼 Strategy & Operations

42- **startup-evaluation** — Startup health diagnosis with entrepreneurship, VC 5T, PMF, runway, team, and next-test frameworks.

43- **anthropic-os** — Self-evolving work method engine.

44- **ai-six-sigma-property-os** — AI + Ontology + DMAIC Black Belt operating model for property work orders, dispatch, quotes, evidence, CTQ dashboards, and MVP quality control.

46## Compatibility

48All skills follow the [Agent Skills](https://agentskills.io) open format. Skills are model-agnostic markdown files compatible with Codex CLI.

50## Skill Contract

52When selecting a skill, read its frontmatter before executing:

54- `assumes` — required operating assumptions.

55- `conflicts_with` — boundaries that must not be silently overridden.

56- `metadata.profile` — `one-shot`, `stateful`, `loop`, or `high-risk` controls.

57- `## Failure Protocol` and `## Output Contract` — standard stop status and receipt.

58- `## Success Metrics` — the minimum observable result for one successful run.

59- `## Quality Gates` — checks that must pass before claiming completion.

61For wiki-writing skills, resolve paths from `system/config.md` when available. Defaults include `SOURCES_DIR=sources/`, `CONCEPTS_DIR=wiki/concepts/`, `ENTITIES_DIR=wiki/entities/`, and `LOG_FILE=system/log.md`.

63## V7 Promotion Gate

65Promote wiki knowledge into a skill, SOP, schema rule, or automation only when:

67- At least two durable wiki/source pages support it, or one high-quality source plus local verification supports it.

68- The rule can be expressed as Trigger -> Execute -> Verify -> State with owner, budget, stop condition, recovery, and write-back.

69- Source immutability, block refs, clipping lifecycle, review queues, permissions, and human approval boundaries are preserved.

70- The change has a cheap check: skill lint, wiki lint, link check, script receipt, test output, dashboard metric, or review receipt.

72## Adoption Ladder

741. Start with `wiki-ingest` + `verify-before-claim`.

752. Add `daily-okr` + `session-learn` after daily ingest and evidence checks are working.

763. Add `cognitive-compile`, `behavior-design`, and `creativity-engine` when the wiki has enough material to synthesize.

774. Add `knowledge-ops` and `loop-engineering` when retrieval or repeated verified execution becomes a bottleneck.

785. Add `graph-engineering` after `loop-engineering` only when explicit dependencies, parallel branches, typed joins, or node-local recovery provide more value than orchestration and review cost.

796. Add `harness-engineering`, `agentic-engineering`, and `agent-teams-command` only when runtime controls, workflow autonomy, or multi-agent process ownership requires them.

81Routing boundary: Loop = temporal depth; Graph = dependency width; Agent Teams = process ownership, IPC, and integration; Harness = runtime scheduler, permissions, and observability.

83## Karpathy LLM OS

85LLM=CPU · Context=RAM · Storage=Disk · Tools=System Calls · Skills=Programs · Harness=Kernel · Agent Teams=Processes

Mark393295827/third-brain-v7-skills · CLAUDE.md

@@ +1 @@

1# Third Brain V7.1 Skills — Claude Code

3For detailed installation, usage, and workflow scenarios, see **[GUIDE.md](GUIDE.md)**.

5You have access to the following Agent Skills. Each skill is a markdown file in `~/.claude/skills/` that defines a specific capability.

7## Skill Categories

9### 📥 Knowledge Pipeline

10- **wiki-ingest** — Ingest sources into an interlinked wiki with source-risk classification, macro-action scope, block refs, clipping archive, Karpathy understanding gate, governance notes, and post-ingest lint. STOW pipeline: Source → Think → Organize → Write.

11- **knowledge-ops** — Multi-layer knowledge management with Markdown-first retrieval, optional ChromaDB vector support, evidence hierarchy, deduplication, and knowledge debt queues.

12- **wiki-lint** — Health-check the wiki across P0/P1 graph health, source refs, frontmatter, links, provenance debt, clipping lifecycle, understanding integrity, and drift.

14### 🔄 Daily Loop

15- **daily-okr** — 7 Key Results daily cycle: Input → Cognition → Wiki → Behavior → Creativity → Output → Feedback, each with evidence.

16- **cognitive-compile** — 8-section deep learning compile: Question → Facts → Concepts → Patterns → Conflicts → Hypotheses → Decision → Action.

18### 🎨 Behavior & Creativity

19- **behavior-design** — Behavior change system: goals → habits → triggers → SOPs → review. Includes HAS (Human Agency Scale).

20- **creativity-engine** — Combinatorial ideation with Lego Building Blocks method. Cross-domain analogies + minimum experiments.

22### 🔬 Research & Quality

23- **deep-research** — STOW-compatible research harness with ChatGPT-style preflight, source/claim ledgers, activity trace, citations, privacy checks, and wiki-ingest handoff.

24- **verify-before-claim** — No completion claims without fresh, scope-matched verification evidence.

26### 🔄 Learning

27- **session-learn** — Extract 7 knowledge signal types from sessions. Closure Protocol for feedback loops.

28- **project-flow-ops** — Triage, plan, track, review across projects.

30### 📊 Context & Cost

31- **context-manager** — Runtime-derived budgets, checkpoints, compaction, retrieval, and capability routing.

32- `token-cost-tracker` — command for estimate/log/report; it is not an Agent Skill.

34### 🏗️ Engineering

35- **loop-engineering** — Temporal-depth control through bounded Goal/Loop/Automation/AutoResearch contracts with state, verification, retry, and recovery.

36- **graph-engineering** — Dependency-width control through bounded static DAGs with explicit dependencies, independent branches, typed joins, and node-local recovery. V7.1 excludes dynamic and cyclic graphs.

37- **agentic-engineering** — Refactor skills and workflows as agent processes with autonomy defaults, delegated-action boundaries, state checkpoints, write-back, and verification gates.

38- **harness-engineering** — Agent runtime kernel: scheduler, permissions, tools as system calls, provenance ledgers, delegated-action gates, observability, recovery, and closed-loop design.

39- **agent-teams-command** — Multi-agent process ownership and orchestration with IPC, async budget envelopes, integration, cleanup, and evidence gates.

41### 💼 Strategy & Operations

42- **startup-evaluation** — Startup health diagnosis with entrepreneurship, VC 5T, PMF, runway, team, and next-test frameworks.

43- **anthropic-os** — Cognitive Symbiont Engine. Livewired + 3B creativity algorithms (Bending/Breaking/Blending). Predictive coding, time-arrow diagnostics.

44- **ai-six-sigma-property-os** — AI + Ontology + DMAIC Black Belt operating model for property work orders, dispatch, quotes, evidence, CTQ dashboards, and MVP quality control.

46## Usage

48Invoke any skill naturally:

49- "Ingest this article into my wiki"

50- "Run a cognitive compile on X"

51- "Create an agent team to build Y"

52- "Launch Anthropic OS for my team" — includes 3B creativity algorithms

53- "Design my Property Agent OS with AI Six Sigma"

54- "Apply 3B to our growth strategy" — Bending/Breaking/Blending

55- "Estimate token cost for this task"

57## Skill Contract

59Before executing a selected skill, read the skill frontmatter and enforce:

61- `assumes`: required operating assumptions for safe use.

62- `conflicts_with`: workflows or assumptions that must not be silently overridden.

63- `metadata.profile`: `one-shot`, `stateful`, `loop`, or `high-risk` control depth.

64- `## Failure Protocol` and `## Output Contract`: standard stop status and receipt.

65- `## Success Metrics`: minimum measurable result for one successful run.

66- `## Quality Gates`: checks required before completion claims.

68For wiki-writing skills, resolve `SOURCES_DIR`, `CONCEPTS_DIR`, `ENTITIES_DIR`, `OUTPUTS_DIR`, and `LOG_FILE` from `system/config.md` when available. If no config exists, use the default STOW layout.

70## Recommended Adoption Ladder

721. Week 1: `wiki-ingest` + `verify-before-claim`.

732. Weeks 2-4: add `daily-okr` + `session-learn`.

743. Month 2+: add `cognitive-compile`, `behavior-design`, and `creativity-engine`.

754. Month 3+: add `knowledge-ops`, `loop-engineering`, `harness-engineering`, and `agentic-engineering`.

765. Dependency scale: add `graph-engineering` after `loop-engineering` only when explicit dependencies, parallel branches, typed joins, or node-local recovery exceed orchestration and review cost.

776. Multi-agent scale: add `agent-teams-command` and `project-flow-ops`.

79## Engineering Routing Boundary

81- Loop Engineering controls temporal depth.

82- Graph Engineering controls dependency width.

83- Agent Teams Command controls process ownership, IPC, and integration.

84- Harness Engineering supplies the runtime scheduler, permissions, and observability.

86A graph node may contain a bounded Loop or Agent Team. V7.1 supports bounded static DAGs, not dynamic or cyclic graphs.

88## Grounding Principles

90### Karpathy LLM OS

91LLM=CPU · Context=RAM · Storage=Disk · Tools=System Calls · Skills=Programs · Harness=Kernel · Agent Teams=Processes

93### 3B Creativity Algorithms

94- **Bending**: Mutate existing success patterns into new contexts

95- **Breaking**: Eliminate worst-performing patterns; break path dependency

96- **Blending**: Fuse elements from different domains for novel patterns

98### Agent Teams

100When the runtime exposes isolated workers, use `agent-teams-command` for admission, ownership, IPC, integration, evidence, and cleanup. Do not bind the durable skill to a product version or model name.

101

@@ −1 +1 @@

1−# Third Brain V7 Skills — Codex CLI

1+# Third Brain V7.1 Skills — Claude Code

3−This repository contains Agent Skills for the Codex CLI environment. Place skills in `~/.agents/skills/`.

3+For detailed installation, usage, and workflow scenarios, see **[GUIDE.md](GUIDE.md)**.

5−V7 treats Obsidian as durable disk and governance, while every skill uses a profile-aware execution contract. V7.1 adds the 20th skill, `graph-engineering`, for bounded static dependency DAGs. Do not turn wiki signals into rules until they pass the promotion gate: repeated source support or local verification, bounded macro action, preserved provenance/permissions, and a cheap objective check.

5+You have access to the following Agent Skills. Each skill is a markdown file in `~/.claude/skills/` that defines a specific capability.

7−## Skills

7+## Skill Categories

99 ### 📥 Knowledge Pipeline

10−- **wiki-ingest** — STOW pipeline with source-risk classification, block refs, clipping lifecycle, understanding gate, promotion controls, and post-ingest lint.

11−- **knowledge-ops** — Multi-layer knowledge management with Markdown-first retrieval, optional vector storage, evidence hierarchy, deduplication, Agent/Wiki flywheel, and knowledge debt queues.

12−- **wiki-lint** — Wiki health check for graph integrity, provenance, links, clipping lifecycle, understanding, and promotion readiness.

10+- **wiki-ingest** — Ingest sources into an interlinked wiki with source-risk classification, macro-action scope, block refs, clipping archive, Karpathy understanding gate, governance notes, and post-ingest lint. STOW pipeline: Source → Think → Organize → Write.

11+- **knowledge-ops** — Multi-layer knowledge management with Markdown-first retrieval, optional ChromaDB vector support, evidence hierarchy, deduplication, and knowledge debt queues.

12+- **wiki-lint** — Health-check the wiki across P0/P1 graph health, source refs, frontmatter, links, provenance debt, clipping lifecycle, understanding integrity, and drift.

1313

1414 ### 🔄 Daily Loop

15−- **daily-okr** — 7 Key Results daily knowledge compound cycle; may consume the scheduled daily knowledge-loop note when present.

16−- **cognitive-compile** — 8-section deep learning compile framework.

15+- **daily-okr** — 7 Key Results daily cycle: Input → Cognition → Wiki → Behavior → Creativity → Output → Feedback, each with evidence.

16+- **cognitive-compile** — 8-section deep learning compile: Question → Facts → Concepts → Patterns → Conflicts → Hypotheses → Decision → Action.

1717

1818 ### 🎨 Behavior & Creativity

19−- **behavior-design** — Behavior change system with HAS framework.

20−- **creativity-engine** — Combinatorial ideation + minimum experiments.

19+- **behavior-design** — Behavior change system: goals → habits → triggers → SOPs → review. Includes HAS (Human Agency Scale).

20+- **creativity-engine** — Combinatorial ideation with Lego Building Blocks method. Cross-domain analogies + minimum experiments.

2121

2222 ### 🔬 Research & Quality

2323 - **deep-research** — STOW-compatible research harness with ChatGPT-style preflight, source/claim ledgers, activity trace, citations, privacy checks, and wiki-ingest handoff.

24−- **verify-before-claim** — Verification-first quality gate.

24+- **verify-before-claim** — No completion claims without fresh, scope-matched verification evidence.

2525

2626 ### 🔄 Learning

27−- **session-learn** — Knowledge extraction with Closure Protocol.

28−- **project-flow-ops** — Project triage and tracking.

27+- **session-learn** — Extract 7 knowledge signal types from sessions. Closure Protocol for feedback loops.

28+- **project-flow-ops** — Triage, plan, track, review across projects.

2929

3030 ### 📊 Context & Cost

31−- **context-manager** — Runtime-derived context budgets, checkpoints, compaction, retrieval, and capability routing.

32−- `token-cost-tracker` is a command under `commands/`, not an Agent Skill.

31+- **context-manager** — Runtime-derived budgets, checkpoints, compaction, retrieval, and capability routing.

32+- `token-cost-tracker` — command for estimate/log/report; it is not an Agent Skill.

3333

3434 ### 🏗️ Engineering

35−- **agentic-engineering** — Agent-as-process workflow refactoring with autonomy defaults, delegated-action boundaries, state checkpoints, write-back, and verification gates.

36−- **loop-engineering** — Temporal-depth control through bounded Trigger -> Execute -> Verify -> State loops, durable contracts, hard budgets, and stop/recovery rules.

37−- **graph-engineering** — Dependency-width control for bounded static DAGs with explicit dependencies, independent branches, typed joins, and node-local recovery; dynamic and cyclic graphs are out of scope in V7.1.

38−- **harness-engineering** — Runtime kernel design: scheduler, permissions, tools as system calls, MCP/Skills/Hooks selection, provenance ledgers, observability, and recovery.

39−- **agent-teams-command** — Multi-agent process ownership, IPC, worktree isolation, async budget envelopes, integration, cleanup, and evidence gates.

35+- **loop-engineering** — Temporal-depth control through bounded Goal/Loop/Automation/AutoResearch contracts with state, verification, retry, and recovery.

36+- **graph-engineering** — Dependency-width control through bounded static DAGs with explicit dependencies, independent branches, typed joins, and node-local recovery. V7.1 excludes dynamic and cyclic graphs.

37+- **agentic-engineering** — Refactor skills and workflows as agent processes with autonomy defaults, delegated-action boundaries, state checkpoints, write-back, and verification gates.

38+- **harness-engineering** — Agent runtime kernel: scheduler, permissions, tools as system calls, provenance ledgers, delegated-action gates, observability, recovery, and closed-loop design.

39+- **agent-teams-command** — Multi-agent process ownership and orchestration with IPC, async budget envelopes, integration, cleanup, and evidence gates.

4040

4141 ### 💼 Strategy & Operations

4242 - **startup-evaluation** — Startup health diagnosis with entrepreneurship, VC 5T, PMF, runway, team, and next-test frameworks.

43−- **anthropic-os** — Self-evolving work method engine.

43+- **anthropic-os** — Cognitive Symbiont Engine. Livewired + 3B creativity algorithms (Bending/Breaking/Blending). Predictive coding, time-arrow diagnostics.

4444 - **ai-six-sigma-property-os** — AI + Ontology + DMAIC Black Belt operating model for property work orders, dispatch, quotes, evidence, CTQ dashboards, and MVP quality control.

4545

46−## Compatibility

46+## Usage

4747

48−All skills follow the [Agent Skills](https://agentskills.io) open format. Skills are model-agnostic markdown files compatible with Codex CLI.

48+Invoke any skill naturally:

49+- "Ingest this article into my wiki"

50+- "Run a cognitive compile on X"

51+- "Create an agent team to build Y"

52+- "Launch Anthropic OS for my team" — includes 3B creativity algorithms

53+- "Design my Property Agent OS with AI Six Sigma"

54+- "Apply 3B to our growth strategy" — Bending/Breaking/Blending

55+- "Estimate token cost for this task"

4956

5057 ## Skill Contract

5158

52−When selecting a skill, read its frontmatter before executing:

59+Before executing a selected skill, read the skill frontmatter and enforce:

5360

54−- `assumes` — required operating assumptions.

55−- `conflicts_with` — boundaries that must not be silently overridden.

56−- `metadata.profile` — `one-shot`, `stateful`, `loop`, or `high-risk` controls.

57−- `## Failure Protocol` and `## Output Contract` — standard stop status and receipt.

58−- `## Success Metrics` — the minimum observable result for one successful run.

59−- `## Quality Gates` — checks that must pass before claiming completion.

61+- `assumes`: required operating assumptions for safe use.

62+- `conflicts_with`: workflows or assumptions that must not be silently overridden.

63+- `metadata.profile`: `one-shot`, `stateful`, `loop`, or `high-risk` control depth.

64+- `## Failure Protocol` and `## Output Contract`: standard stop status and receipt.

65+- `## Success Metrics`: minimum measurable result for one successful run.

66+- `## Quality Gates`: checks required before completion claims.

6067

61−For wiki-writing skills, resolve paths from `system/config.md` when available. Defaults include `SOURCES_DIR=sources/`, `CONCEPTS_DIR=wiki/concepts/`, `ENTITIES_DIR=wiki/entities/`, and `LOG_FILE=system/log.md`.

68+For wiki-writing skills, resolve `SOURCES_DIR`, `CONCEPTS_DIR`, `ENTITIES_DIR`, `OUTPUTS_DIR`, and `LOG_FILE` from `system/config.md` when available. If no config exists, use the default STOW layout.

6269

63−## V7 Promotion Gate

70+## Recommended Adoption Ladder

6471

65−Promote wiki knowledge into a skill, SOP, schema rule, or automation only when:

72+1. Week 1: `wiki-ingest` + `verify-before-claim`.

73+2. Weeks 2-4: add `daily-okr` + `session-learn`.

74+3. Month 2+: add `cognitive-compile`, `behavior-design`, and `creativity-engine`.

75+4. Month 3+: add `knowledge-ops`, `loop-engineering`, `harness-engineering`, and `agentic-engineering`.

76+5. Dependency scale: add `graph-engineering` after `loop-engineering` only when explicit dependencies, parallel branches, typed joins, or node-local recovery exceed orchestration and review cost.

77+6. Multi-agent scale: add `agent-teams-command` and `project-flow-ops`.

6678

67−- At least two durable wiki/source pages support it, or one high-quality source plus local verification supports it.

68−- The rule can be expressed as Trigger -> Execute -> Verify -> State with owner, budget, stop condition, recovery, and write-back.

69−- Source immutability, block refs, clipping lifecycle, review queues, permissions, and human approval boundaries are preserved.

70−- The change has a cheap check: skill lint, wiki lint, link check, script receipt, test output, dashboard metric, or review receipt.

79+## Engineering Routing Boundary

7180

72−## Adoption Ladder

81+- Loop Engineering controls temporal depth.

82+- Graph Engineering controls dependency width.

83+- Agent Teams Command controls process ownership, IPC, and integration.

84+- Harness Engineering supplies the runtime scheduler, permissions, and observability.

7385

74−1. Start with `wiki-ingest` + `verify-before-claim`.

75−2. Add `daily-okr` + `session-learn` after daily ingest and evidence checks are working.

76−3. Add `cognitive-compile`, `behavior-design`, and `creativity-engine` when the wiki has enough material to synthesize.

77−4. Add `knowledge-ops` and `loop-engineering` when retrieval or repeated verified execution becomes a bottleneck.

78−5. Add `graph-engineering` after `loop-engineering` only when explicit dependencies, parallel branches, typed joins, or node-local recovery provide more value than orchestration and review cost.

79−6. Add `harness-engineering`, `agentic-engineering`, and `agent-teams-command` only when runtime controls, workflow autonomy, or multi-agent process ownership requires them.

86+A graph node may contain a bounded Loop or Agent Team. V7.1 supports bounded static DAGs, not dynamic or cyclic graphs.

8087

81−Routing boundary: Loop = temporal depth; Graph = dependency width; Agent Teams = process ownership, IPC, and integration; Harness = runtime scheduler, permissions, and observability.

88+## Grounding Principles

8289

83−## Karpathy LLM OS

84−

90+### Karpathy LLM OS

8591 LLM=CPU · Context=RAM · Storage=Disk · Tools=System Calls · Skills=Programs · Harness=Kernel · Agent Teams=Processes

92+

93+### 3B Creativity Algorithms

94+- **Bending**: Mutate existing success patterns into new contexts

95+- **Breaking**: Eliminate worst-performing patterns; break path dependency

96+- **Blending**: Fuse elements from different domains for novel patterns

97+

98+### Agent Teams

99+

100+When the runtime exposes isolated workers, use `agent-teams-command` for admission, ownership, IPC, integration, evidence, and cleanup. Do not bind the durable skill to a product version or model name.

86101