ai-collab-open-system 0.1.0

package/.aict/examples/content-production-harvest/CASE.md

@@ -0,0 +1,87 @@
+# Content production and harvest
+
+This is a fully synthetic case. It does not contain private customer material, real raw conversations, local paths, or private operational routes.
+
+## Confusing raw input
+
+A creator has notes for a launch essay, a short post, and a newsletter, but every AI draft sounds generic and the useful lines are lost after revisions.
+
+## Likely single-agent failure
+
+A likely single-agent answer would produce a fluent artifact from the messy request, but it would not preserve explicit context, acceptance, guard review, handoff state, and harvest learning.
+
+## AI Collaboration OS process
+
+1. Context package: Profile: wants concrete language, low hype, and examples before claims. Context: synthetic launch of a local recipe planner for busy families.
+2. Acceptance card: Done means one essay outline, one short post, one reused line bank, and a review note identifying generic claims to remove.
+3. Execution prompt: Draft from the context package. Keep claims tied to the synthetic case. Produce a line bank separately from the essay.
+4. Guard review: Guard rejects two abstract claims and asks for one concrete household planning scene before approving.
+5. Handoff note: Draft approved after replacing abstract claims. Pending: choose which harvested lines become future reusable prompts.
+6. Harvest seed: Reusable lines: 'Do not make the assistant sound smarter than the user's evidence' and 'One vivid case before one broad claim.'
+
+## Context package
+
+Profile: wants concrete language, low hype, and examples before claims. Context: synthetic launch of a local recipe planner for busy families.
+
+## Acceptance card
+
+Done means one essay outline, one short post, one reused line bank, and a review note identifying generic claims to remove.
+
+## Guard review
+
+Guard rejects two abstract claims and asks for one concrete household planning scene before approving.
+
+## Handoff note
+
+Draft approved after replacing abstract claims. Pending: choose which harvested lines become future reusable prompts.
+
+## Harvest seed
+
+Reusable lines: 'Do not make the assistant sound smarter than the user's evidence' and 'One vivid case before one broad claim.'
+
+## Before/after comparison
+
+A raw chat would produce a polished draft and lose the reusable voice rules. The workspace extracts future writing guidance.
+
+## Messy starting point
+
+A creator has notes for a launch essay, a short post, and a newsletter, but every AI draft sounds generic and the useful lines are lost after revisions.
+
+## Workspace setup
+
+Use profile for voice constraints, context for audience and source notes, acceptance for what a usable draft must do, guard for anti-generic review, and harvest for reusable lines.
+
+## Profile/context
+
+Profile: wants concrete language, low hype, and examples before claims. Context: synthetic launch of a local recipe planner for busy families.
+
+## Acceptance
+
+Done means one essay outline, one short post, one reused line bank, and a review note identifying generic claims to remove.
+
+## Execution prompt
+
+```text
+Draft from the context package. Keep claims tied to the synthetic case. Produce a line bank separately from the essay.
+```
+
+## Handoff
+
+Draft approved after replacing abstract claims. Pending: choose which harvested lines become future reusable prompts.
+
+## Harvest
+
+Reusable lines: 'Do not make the assistant sound smarter than the user's evidence' and 'One vivid case before one broad claim.'
+
+## What changes compared with a single raw AI chat
+
+A raw chat would produce a polished draft and lose the reusable voice rules. The workspace extracts future writing guidance.
+
+## Artifacts
+
+- `artifacts/context-package.md`
+- `artifacts/acceptance-card.md`
+- `artifacts/execution-prompt.md`
+- `artifacts/guard-review.md`
+- `artifacts/handoff-note.md`
+- `artifacts/harvest-seed.md`

package/.aict/examples/content-production-harvest/artifacts/acceptance-card.md

@@ -0,0 +1,28 @@
+# Acceptance card - Content production and harvest
+
+## Source case
+
+- Case id: `content-production-harvest`
+- Case title: Content production and harvest
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this before implementation, drafting, research, or judgment work. Ask the assistant to treat these criteria as the pass/fail surface.
+
+## Synthetic content
+
+Done means one essay outline, one short post, one reused line bank, and a review note identifying generic claims to remove.
+
+## Review note
+
+Reject work that claims completion without evidence tied to this card.
+
+## Next step
+
+Use this card with the execution prompt and later guard review.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/content-production-harvest/artifacts/context-package.md

@@ -0,0 +1,28 @@
+# Context package - Content production and harvest
+
+## Source case
+
+- Case id: `content-production-harvest`
+- Case title: Content production and harvest
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this before asking an AI tool to continue the task. It gives the assistant the working style, scope, constraints, and known evidence.
+
+## Synthetic content
+
+Profile: wants concrete language, low hype, and examples before claims. Context: synthetic launch of a local recipe planner for busy families.
+
+## Review note
+
+Check that facts and assumptions are separated before execution starts.
+
+## Next step
+
+Use this context to write or verify the acceptance card.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/content-production-harvest/artifacts/execution-prompt.md

@@ -0,0 +1,30 @@
+# Execution prompt - Content production and harvest
+
+## Source case
+
+- Case id: `content-production-harvest`
+- Case title: Content production and harvest
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this into the selected AI tool after the context package and acceptance card.
+
+## Synthetic content
+
+```text
+Draft from the context package. Keep claims tied to the synthetic case. Produce a line bank separately from the essay.
+```
+
+## Review note
+
+Confirm the prompt does not expand scope beyond acceptance.
+
+## Next step
+
+Run guard review on the first artifact produced from this prompt.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/content-production-harvest/artifacts/guard-review.md

@@ -0,0 +1,28 @@
+# Guard review - Content production and harvest
+
+## Source case
+
+- Case id: `content-production-harvest`
+- Case title: Content production and harvest
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Use this as the review stance after the first artifact exists. It challenges evidence, privacy, scope, and acceptance alignment.
+
+## Synthetic content
+
+Guard rejects two abstract claims and asks for one concrete household planning scene before approving.
+
+## Review note
+
+A review is not a pass unless it names evidence and residual risk.
+
+## Next step
+
+Fix any blocking finding, then write a handoff note.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/content-production-harvest/artifacts/handoff-note.md

@@ -0,0 +1,28 @@
+# Handoff note - Content production and harvest
+
+## Source case
+
+- Case id: `content-production-harvest`
+- Case title: Content production and harvest
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this into the next session or tool so work resumes from current state instead of restarting.
+
+## Synthetic content
+
+Draft approved after replacing abstract claims. Pending: choose which harvested lines become future reusable prompts.
+
+## Review note
+
+Check that completed, pending, blocked, and next action are distinguishable.
+
+## Next step
+
+Use the handoff note as input to harvest.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/content-production-harvest/artifacts/harvest-seed.md

@@ -0,0 +1,28 @@
+# Harvest seed - Content production and harvest
+
+## Source case
+
+- Case id: `content-production-harvest`
+- Case title: Content production and harvest
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Save this after the loop to preserve reusable knowledge without copying private raw material.
+
+## Synthetic content
+
+Reusable lines: 'Do not make the assistant sound smarter than the user's evidence' and 'One vivid case before one broad claim.'
+
+## Review note
+
+Do not generalize from the synthetic case unless the pattern appears in future work.
+
+## Next step
+
+Move reusable prompts or rules into the appropriate workspace file.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/multi-tool-collaboration/CASE.md

@@ -0,0 +1,87 @@
+# Multi-tool collaboration
+
+This is a fully synthetic case. It does not contain private customer material, real raw conversations, local paths, or private operational routes.
+
+## Confusing raw input
+
+A user starts planning in one assistant, implements in another, and reviews in a third. Each tool uses a different memory of the task.
+
+## Likely single-agent failure
+
+A likely single-agent answer would produce a fluent artifact from the messy request, but it would not preserve explicit context, acceptance, guard review, handoff state, and harvest learning.
+
+## AI Collaboration OS process
+
+1. Context package: Profile: wants one controller view and concise cross-tool handoffs. Context: synthetic static-site cleanup spread across Codex, Claude Code, and Cursor.
+2. Acceptance card: Done means each tool reads the same core contract, uses the same acceptance card, and leaves a handoff note with changed artifacts.
+3. Execution prompt: Tool A packages context, Tool B edits the synthetic site copy, Tool C reviews against acceptance. All tools must cite the same shared contract.
+4. Guard review: Guard catches that Cursor-specific instructions duplicated rules instead of pointing to the shared contract.
+5. Handoff note: Fix adapter drift by replacing duplicated rules with a pointer to SHARED_CORE_CONTRACT.md.
+6. Harvest seed: Reusable pattern: adapters must be thin pointers, not six separate rule systems.
+
+## Context package
+
+Profile: wants one controller view and concise cross-tool handoffs. Context: synthetic static-site cleanup spread across Codex, Claude Code, and Cursor.
+
+## Acceptance card
+
+Done means each tool reads the same core contract, uses the same acceptance card, and leaves a handoff note with changed artifacts.
+
+## Guard review
+
+Guard catches that Cursor-specific instructions duplicated rules instead of pointing to the shared contract.
+
+## Handoff note
+
+Fix adapter drift by replacing duplicated rules with a pointer to SHARED_CORE_CONTRACT.md.
+
+## Harvest seed
+
+Reusable pattern: adapters must be thin pointers, not six separate rule systems.
+
+## Before/after comparison
+
+A raw multi-tool workflow creates rule drift. The workspace keeps a shared contract and thin adapters.
+
+## Messy starting point
+
+A user starts planning in one assistant, implements in another, and reviews in a third. Each tool uses a different memory of the task.
+
+## Workspace setup
+
+Use the adapter files to point each tool to the same shared core contract and the same context, acceptance, guard, handoff, and harvest files.
+
+## Profile/context
+
+Profile: wants one controller view and concise cross-tool handoffs. Context: synthetic static-site cleanup spread across Codex, Claude Code, and Cursor.
+
+## Acceptance
+
+Done means each tool reads the same core contract, uses the same acceptance card, and leaves a handoff note with changed artifacts.
+
+## Execution prompt
+
+```text
+Tool A packages context, Tool B edits the synthetic site copy, Tool C reviews against acceptance. All tools must cite the same shared contract.
+```
+
+## Handoff
+
+Fix adapter drift by replacing duplicated rules with a pointer to SHARED_CORE_CONTRACT.md.
+
+## Harvest
+
+Reusable pattern: adapters must be thin pointers, not six separate rule systems.
+
+## What changes compared with a single raw AI chat
+
+A raw multi-tool workflow creates rule drift. The workspace keeps a shared contract and thin adapters.
+
+## Artifacts
+
+- `artifacts/context-package.md`
+- `artifacts/acceptance-card.md`
+- `artifacts/execution-prompt.md`
+- `artifacts/guard-review.md`
+- `artifacts/handoff-note.md`
+- `artifacts/harvest-seed.md`

package/.aict/examples/multi-tool-collaboration/artifacts/acceptance-card.md

@@ -0,0 +1,28 @@
+# Acceptance card - Multi-tool collaboration
+
+## Source case
+
+- Case id: `multi-tool-collaboration`
+- Case title: Multi-tool collaboration
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this before implementation, drafting, research, or judgment work. Ask the assistant to treat these criteria as the pass/fail surface.
+
+## Synthetic content
+
+Done means each tool reads the same core contract, uses the same acceptance card, and leaves a handoff note with changed artifacts.
+
+## Review note
+
+Reject work that claims completion without evidence tied to this card.
+
+## Next step
+
+Use this card with the execution prompt and later guard review.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/multi-tool-collaboration/artifacts/context-package.md

@@ -0,0 +1,28 @@
+# Context package - Multi-tool collaboration
+
+## Source case
+
+- Case id: `multi-tool-collaboration`
+- Case title: Multi-tool collaboration
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this before asking an AI tool to continue the task. It gives the assistant the working style, scope, constraints, and known evidence.
+
+## Synthetic content
+
+Profile: wants one controller view and concise cross-tool handoffs. Context: synthetic static-site cleanup spread across Codex, Claude Code, and Cursor.
+
+## Review note
+
+Check that facts and assumptions are separated before execution starts.
+
+## Next step
+
+Use this context to write or verify the acceptance card.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/multi-tool-collaboration/artifacts/execution-prompt.md

@@ -0,0 +1,30 @@
+# Execution prompt - Multi-tool collaboration
+
+## Source case
+
+- Case id: `multi-tool-collaboration`
+- Case title: Multi-tool collaboration
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this into the selected AI tool after the context package and acceptance card.
+
+## Synthetic content
+
+```text
+Tool A packages context, Tool B edits the synthetic site copy, Tool C reviews against acceptance. All tools must cite the same shared contract.
+```
+
+## Review note
+
+Confirm the prompt does not expand scope beyond acceptance.
+
+## Next step
+
+Run guard review on the first artifact produced from this prompt.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/multi-tool-collaboration/artifacts/guard-review.md

@@ -0,0 +1,28 @@
+# Guard review - Multi-tool collaboration
+
+## Source case
+
+- Case id: `multi-tool-collaboration`
+- Case title: Multi-tool collaboration
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Use this as the review stance after the first artifact exists. It challenges evidence, privacy, scope, and acceptance alignment.
+
+## Synthetic content
+
+Guard catches that Cursor-specific instructions duplicated rules instead of pointing to the shared contract.
+
+## Review note
+
+A review is not a pass unless it names evidence and residual risk.
+
+## Next step
+
+Fix any blocking finding, then write a handoff note.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/multi-tool-collaboration/artifacts/handoff-note.md

@@ -0,0 +1,28 @@
+# Handoff note - Multi-tool collaboration
+
+## Source case
+
+- Case id: `multi-tool-collaboration`
+- Case title: Multi-tool collaboration
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this into the next session or tool so work resumes from current state instead of restarting.
+
+## Synthetic content
+
+Fix adapter drift by replacing duplicated rules with a pointer to SHARED_CORE_CONTRACT.md.
+
+## Review note
+
+Check that completed, pending, blocked, and next action are distinguishable.
+
+## Next step
+
+Use the handoff note as input to harvest.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/multi-tool-collaboration/artifacts/harvest-seed.md

@@ -0,0 +1,28 @@
+# Harvest seed - Multi-tool collaboration
+
+## Source case
+
+- Case id: `multi-tool-collaboration`
+- Case title: Multi-tool collaboration
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Save this after the loop to preserve reusable knowledge without copying private raw material.
+
+## Synthetic content
+
+Reusable pattern: adapters must be thin pointers, not six separate rule systems.
+
+## Review note
+
+Do not generalize from the synthetic case unless the pattern appears in future work.
+
+## Next step
+
+Move reusable prompts or rules into the appropriate workspace file.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/personal-judgment-growth-assistant/CASE.md

@@ -0,0 +1,87 @@
+# Personal judgment / growth assistant
+
+This is a fully synthetic case. It does not contain private customer material, real raw conversations, local paths, or private operational routes.
+
+## Confusing raw input
+
+A person asks AI to help reflect on a difficult career choice, but the assistant overreaches, sounds certain, and treats a private decision as optimization.
+
+## Likely single-agent failure
+
+A likely single-agent answer would produce a fluent artifact from the messy request, but it would not preserve explicit context, acceptance, guard review, handoff state, and harvest learning.
+
+## AI Collaboration OS process
+
+1. Context package: Profile: wants questions that clarify values, not commands. Context: synthetic choice between a stable internal role and a risky independent project.
+2. Acceptance card: Done means the assistant lists options, tradeoffs, unknowns, and decision questions without claiming to know what the user should do.
+3. Execution prompt: Help structure the decision. Do not decide for the user. Separate values, constraints, fears, evidence, and next reversible step.
+4. Guard review: Guard rejects one sentence that says the user should choose the risky project and replaces it with a decision question.
+5. Handoff note: Current state: decision map complete. Pending: user answers three value questions offline.
+6. Harvest seed: Reusable pattern: judgment assistance must preserve human agency and make uncertainty explicit.
+
+## Context package
+
+Profile: wants questions that clarify values, not commands. Context: synthetic choice between a stable internal role and a risky independent project.
+
+## Acceptance card
+
+Done means the assistant lists options, tradeoffs, unknowns, and decision questions without claiming to know what the user should do.
+
+## Guard review
+
+Guard rejects one sentence that says the user should choose the risky project and replaces it with a decision question.
+
+## Handoff note
+
+Current state: decision map complete. Pending: user answers three value questions offline.
+
+## Harvest seed
+
+Reusable pattern: judgment assistance must preserve human agency and make uncertainty explicit.
+
+## Before/after comparison
+
+A raw chat may sound confident and directive. The workspace keeps human judgment primary.
+
+## Messy starting point
+
+A person asks AI to help reflect on a difficult career choice, but the assistant overreaches, sounds certain, and treats a private decision as optimization.
+
+## Workspace setup
+
+Use profile to define decision boundaries, context to state the situation without private identifiers, acceptance to require options and tradeoffs, guard to prevent overclaiming, and harvest to save decision questions.
+
+## Profile/context
+
+Profile: wants questions that clarify values, not commands. Context: synthetic choice between a stable internal role and a risky independent project.
+
+## Acceptance
+
+Done means the assistant lists options, tradeoffs, unknowns, and decision questions without claiming to know what the user should do.
+
+## Execution prompt
+
+```text
+Help structure the decision. Do not decide for the user. Separate values, constraints, fears, evidence, and next reversible step.
+```
+
+## Handoff
+
+Current state: decision map complete. Pending: user answers three value questions offline.
+
+## Harvest
+
+Reusable pattern: judgment assistance must preserve human agency and make uncertainty explicit.
+
+## What changes compared with a single raw AI chat
+
+A raw chat may sound confident and directive. The workspace keeps human judgment primary.
+
+## Artifacts
+
+- `artifacts/context-package.md`
+- `artifacts/acceptance-card.md`
+- `artifacts/execution-prompt.md`
+- `artifacts/guard-review.md`
+- `artifacts/handoff-note.md`
+- `artifacts/harvest-seed.md`

package/.aict/examples/personal-judgment-growth-assistant/artifacts/acceptance-card.md

@@ -0,0 +1,28 @@
+# Acceptance card - Personal judgment / growth assistant
+
+## Source case
+
+- Case id: `personal-judgment-growth-assistant`
+- Case title: Personal judgment / growth assistant
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this before implementation, drafting, research, or judgment work. Ask the assistant to treat these criteria as the pass/fail surface.
+
+## Synthetic content
+
+Done means the assistant lists options, tradeoffs, unknowns, and decision questions without claiming to know what the user should do.
+
+## Review note
+
+Reject work that claims completion without evidence tied to this card.
+
+## Next step
+
+Use this card with the execution prompt and later guard review.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.

package/.aict/examples/personal-judgment-growth-assistant/artifacts/context-package.md

@@ -0,0 +1,28 @@
+# Context package - Personal judgment / growth assistant
+
+## Source case
+
+- Case id: `personal-judgment-growth-assistant`
+- Case title: Personal judgment / growth assistant
+- Privacy status: fully synthetic
+- Private material: none
+
+## How to use
+
+Paste this before asking an AI tool to continue the task. It gives the assistant the working style, scope, constraints, and known evidence.
+
+## Synthetic content
+
+Profile: wants questions that clarify values, not commands. Context: synthetic choice between a stable internal role and a risky independent project.
+
+## Review note
+
+Check that facts and assumptions are separated before execution starts.
+
+## Next step
+
+Use this context to write or verify the acceptance card.
+
+## Why this exists
+
+This artifact makes the case runnable and reviewable. A raw chat can produce a smooth answer, but this file preserves the specific state needed for profile, context, acceptance, guard, handoff, and harvest work.