docs(ecosystem): track phase3 audited artifact flow

This commit is contained in:
Saravana Achu Mac 2026-04-03 19:49:06 -07:00
parent e6b58b7187
commit 428112b702
8 changed files with 246 additions and 6 deletions

View File

@ -209,6 +209,7 @@ Commits:
- `41fa2cd` drafted the event taxonomy
- `76f1b47` added canonical Phase 1 event schemas and example payloads
- `78918fb` extended canonical artifact payloads so Phase 2 can reuse `artifact.created` and `artifact.linked` for `plan`, `routine`, and `habit`
- `e6b58b7` extended canonical artifact payloads so Phase 3 can reuse `artifact.created` for `trail-report` and preserve the audited note -> memory seam without introducing a second event family
---

View File

@ -145,7 +145,7 @@ Example:
- ChronoMind imports the latest plan into a routine artifact and preserves plan-created causation
- EffoRise imports the latest routine into a habit artifact and preserves routine-linked causation
- Codex completed a manual runtime walkthrough on `2026-04-03` using `/tmp/bytelyst-phase2-walkthrough.WpEjZu`, which produced the causal chain `evt_plan_created_b202f23f69a4 -> evt_routine_created_4c07d3fdcd07 -> evt_routine_linked_08b9db8a7013 -> evt_habit_created_f532f13332d1 -> evt_habit_linked_7be9a47928e3`
- [ ] Cowork -> ActionTrail -> NoteLett -> MindLyst audited artifact flow
- [x] Cowork -> ActionTrail -> NoteLett -> MindLyst audited artifact flow
Repos:
- `oss/learning_ai_claw-cowork`
- `learning_ai_trails`
@ -153,7 +153,16 @@ Example:
- `learning_multimodal_memory_agents`
- `learning_ai_common_plat`
Commits:
- pending
- `e6b58b7`
- `e294cf5`
- `7678c9b`
- `43819f1`
Status note:
- shared contract now supports a canonical `trail-report` payload while reusing generic `artifact.created`, `artifact.linked`, and `memory.entry.created`
- ActionTrail imports Cowork audit-export JSON into a persisted `trail-report` artifact plus latest created-event index
- NoteLett imports the latest `trail-report` into a markdown audit note and preserves upstream trail-report causation
- MindLyst imports the latest audit note into an `insight` memory proposal and preserves upstream note-link lineage
- Codex completed a manual runtime walkthrough on `2026-04-03` using `/tmp/bytelyst-phase3-walkthrough.myuu1z`, which produced the causal chain `evt_trail_created_b40197d5302e -> evt_note_created_abc1ad821974 -> evt_note_linked_9fd9a2b03363 -> evt_memory_8103b59fb40c`
---
@ -189,3 +198,4 @@ Execution status:
- Windsurf verified the `learning_ai_notes` Phase 1 backend path with `cd backend && npx vitest run`; 22 files passed and 106 tests passed, including the focused ecosystem-phase1 coverage
- shared contract now also supports the Phase 2 `plan -> routine -> habit` seam through `artifact.created` and `artifact.linked`
- Phase 2 walkthrough is complete and tracked in [`PHASE2_PLAN_ROUTINE_HABIT_EXECUTION_PLAN.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/PHASE2_PLAN_ROUTINE_HABIT_EXECUTION_PLAN.md)
- Phase 3 walkthrough is complete and tracked in [`PHASE3_COWORK_ACTIONTRAIL_NOTE_MEMORY_EXECUTION_PLAN.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/PHASE3_COWORK_ACTIONTRAIL_NOTE_MEMORY_EXECUTION_PLAN.md)

View File

@ -1,6 +1,6 @@
# Ecosystem Shared Artifact Schema
> **Status:** Phase 2 baseline implemented
> **Status:** Phase 3 baseline implemented
> **Owner:** `learning_ai_common_plat`
> **Purpose:** Define the canonical cross-product artifact model so notes, transcripts, plans, memories, trails, routes, and agent outputs can interoperate cleanly.
@ -190,7 +190,7 @@ Phase 2 adopters:
- [ ] define migration strategy for first adopters
- [x] add examples for LysnrAI, NoteLett, and MindLyst
- [x] add contract-backed examples for FlowMonk, ChronoMind, and EffoRise
- [ ] add examples for Cowork and ActionTrail
- [x] add examples for Cowork and ActionTrail
Current implementation location:
@ -202,3 +202,4 @@ Commits:
- `eae3409` drafted the initial stub
- `76f1b47` added Phase 1 artifact schemas and canonical fixtures
- `78918fb` added Phase 2 artifact schemas for `plan`, `routine`, and `habit`
- `e6b58b7` added Phase 3 `trail-report` schema coverage and examples for Cowork/ActionTrail

View File

@ -0,0 +1,128 @@
# Phase 3 Execution Plan
> **Flow:** `Cowork -> ActionTrail -> NoteLett -> MindLyst`
> **Status:** Implemented baseline
> **Owner:** `Codex`
> **Purpose:** Standardize the audited artifact flow where Cowork audit exports become ActionTrail trail reports, then readable NoteLett notes, then MindLyst insight memories.
---
## 1. Scope
Phase 3 proves that a Cowork execution can create durable downstream value without requiring a live Rust/Tauri integration in the same implementation window.
Baseline seam:
1. Cowork exports structured audit JSON.
2. ActionTrail imports that export into a canonical `trail-report` artifact.
3. NoteLett imports the latest `trail-report` into a readable markdown note.
4. MindLyst imports the latest audit note into an `insight` memory proposal.
This phase intentionally treats Cowork's existing audit export as the producer contract. It does not require new desktop runtime plumbing.
---
## 2. Repos In Scope
- `oss/learning_ai_claw-cowork`
- `learning_ai_trails`
- `learning_ai_notes`
- `learning_multimodal_memory_agents`
- `learning_ai_common_plat`
---
## 3. Implementation Decisions
- Reuse generic ecosystem events:
- `artifact.created`
- `artifact.linked`
- `memory.entry.created`
- Add a canonical `trail-report` payload to the shared artifact schema.
- Keep ActionTrail's Phase 3 seam as a pure module so it can be tested without booting the backend.
- Reuse the existing NoteLett and MindLyst import pattern from Phase 1.
- Preserve causal lineage across all persisted latest-event indexes.
---
## 4. Code Surfaces
### `learning_ai_common_plat`
- `packages/events/src/ecosystem.ts`
- `packages/events/src/ecosystem.test.ts`
### `learning_ai_trails`
- `backend/src/lib/ecosystem-phase3.ts`
- `backend/src/lib/ecosystem-phase3.test.ts`
### `learning_ai_notes`
- `backend/src/lib/ecosystem-phase3.ts`
- `backend/src/lib/ecosystem-phase3.test.ts`
### `learning_multimodal_memory_agents`
- `mindlyst-native/web/src/lib/ecosystem-phase3.ts`
- `mindlyst-native/web/src/lib/ecosystem-phase3.test.ts`
---
## 5. Verification Commands
- `cd learning_ai_common_plat/packages/events && pnpm exec vitest run src/ecosystem.test.ts`
- `cd learning_ai_trails && node --experimental-strip-types --test backend/src/lib/ecosystem-phase3.test.ts`
- `cd learning_ai_notes && node --experimental-strip-types --test backend/src/lib/ecosystem-phase3.test.ts`
- `cd learning_multimodal_memory_agents/mindlyst-native/web && node --experimental-strip-types --test src/lib/ecosystem-phase3.test.ts`
---
## 6. Manual Walkthrough Evidence
Walkthrough root:
- `/tmp/bytelyst-phase3-walkthrough.myuu1z`
Observed causal chain:
- `evt_trail_created_b40197d5302e`
- `evt_note_created_abc1ad821974`
- `evt_note_linked_9fd9a2b03363`
- `evt_memory_8103b59fb40c`
Observed artifact chain:
- `art_trail_5af2de023e77`
- `art_note_2146a3d0cd11`
- `art_memory_11f691d1b201`
Observed semantics:
- Cowork audit export produced `4` audited actions and `2` safety signals.
- ActionTrail created a `trail-report` artifact with shared correlation ID `corr_e735382a77d6`.
- NoteLett created a markdown audit note linked back to the `trail-report`.
- MindLyst created an `insight` memory with source artifact IDs for both the note and the trail report.
---
## 7. Done Criteria
- [x] shared contract accepts `trail-report` artifacts
- [x] ActionTrail can persist a canonical `trail-report` artifact and latest created-event index
- [x] NoteLett can persist a canonical note artifact derived from the latest `trail-report`
- [x] MindLyst can persist a canonical `insight` memory derived from the latest audit note
- [x] all focused module tests pass
- [x] one manual walkthrough proves the end-to-end causal chain
- [ ] repo-local runtime route wiring exists for all three products
The remaining unchecked item is explicitly deferred. Phase 3 baseline is still considered complete because the artifact seam is implemented and verified.
---
## 8. Commits
- `e6b58b7` shared Phase 3 contract support for `trail-report`
- `e294cf5` ActionTrail `trail-report` producer seam
- `7678c9b` NoteLett audited note import seam
- `43819f1` MindLyst audited memory import seam

View File

@ -27,6 +27,7 @@ Recommended structure:
- [`ECOSYSTEM_PERSONAL_TIMELINE_PRD.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/ECOSYSTEM_PERSONAL_TIMELINE_PRD.md) — unified user activity stream
- [`PHASE1_TRANSCRIPT_NOTE_MEMORY_EXECUTION_PLAN.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/PHASE1_TRANSCRIPT_NOTE_MEMORY_EXECUTION_PLAN.md) — execution-ready plan for the first golden flow
- [`PHASE2_PLAN_ROUTINE_HABIT_EXECUTION_PLAN.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/PHASE2_PLAN_ROUTINE_HABIT_EXECUTION_PLAN.md) — execution-ready plan for the second golden flow
- [`PHASE3_COWORK_ACTIONTRAIL_NOTE_MEMORY_EXECUTION_PLAN.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/PHASE3_COWORK_ACTIONTRAIL_NOTE_MEMORY_EXECUTION_PLAN.md) — execution-ready plan for the audited artifact flow
- [`delegation/`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/delegation) — ready-to-send task packs and launcher lines for Saravana, Codex, Windsurf, and Cursor
- [`adoption/`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/adoption) — per-product adoption notes
@ -49,4 +50,4 @@ Tracking convention:
Current implementation package:
- [`packages/events`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/packages/events) holds the first concrete ecosystem contract code and now covers both the Phase 1 and Phase 2 artifact/event seams.
- [`packages/events`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/packages/events) holds the first concrete ecosystem contract code and now covers the Phase 1, Phase 2, and Phase 3 artifact/event seams.

View File

@ -10,7 +10,7 @@ Recommended file pattern:
- `learning_ai_notes.md`
- `learning_ai_flowmonk.md`
- `learning_ai_trails.md`
- `oss_learning_ai_claw_cowork.md`
- `learning_ai_claw-cowork.md`
Current docs:
@ -20,6 +20,8 @@ Current docs:
- [`learning_ai_flowmonk.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/adoption/learning_ai_flowmonk.md)
- [`learning_ai_clock.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/adoption/learning_ai_clock.md)
- [`learning_ai_efforise.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/adoption/learning_ai_efforise.md)
- [`learning_ai_trails.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/adoption/learning_ai_trails.md)
- [`learning_ai_claw-cowork.md`](/Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/adoption/learning_ai_claw-cowork.md)
Each adoption doc should cover:

View File

@ -0,0 +1,47 @@
# Claw Cowork Ecosystem Adoption
> **Repo:** `oss/learning_ai_claw-cowork`
> **Ecosystem focus:** audited artifact producer for Phase 3
> **Status:** existing producer surface reused
---
## 1. Role In The Ecosystem
Claw Cowork is the audit-producing source system for Phase 3.
It already provides:
- structured append-only audit logs
- filtered audit queries
- JSON export
- CSV export
- desktop audit inspection UI
Those existing surfaces make Cowork the right producer for the audited artifact flow without adding risky new desktop plumbing.
---
## 2. Current Adoption
Phase 3 baseline reuses the existing desktop export surface:
- `query_audit_log`
- `export_audit_json`
- `export_audit_csv`
ActionTrail now treats the exported JSON audit shape as the canonical import input for `trail-report` creation.
---
## 3. Remaining Follow-Up
- add an explicit product-local helper for exporting ecosystem-ready audit payloads
- optionally emit canonical runtime events alongside the current local audit logs
- add signed dispatch or upload flow from Cowork desktop into ActionTrail when runtime integration is prioritized
---
## 4. Commits
- existing producer surface predates Phase 3 baseline

View File

@ -0,0 +1,50 @@
# ActionTrail Ecosystem Adoption
> **Repo:** `learning_ai_trails`
> **Ecosystem focus:** Cowork audit -> trail-report artifact golden flow
> **Status:** Phase 3 baseline implemented
---
## 1. Role In The Ecosystem
ActionTrail is the canonical audit-normalization layer for Phase 3.
It should:
- accept structured Cowork audit exports
- convert them into durable ecosystem `trail-report` artifacts
- preserve causation and correlation for downstream consumers
- make the audit flow readable without requiring Cowork runtime knowledge in every consumer
---
## 2. Current Adoption
- `backend/src/lib/ecosystem-phase3.ts` creates and persists a canonical `trail-report`
- latest indexes now include:
- `latest-trail-report.json`
- `latest-trail-report-created-event.json`
- the Phase 3 payload captures:
- action counts
- tool call counts
- approval counts
- failure counts
- safety signal counts
- task list
- action breakdown
- normalized entry list
---
## 3. Remaining Follow-Up
- add optional route/service wiring for importing Cowork exports through the live backend
- map ActionTrail-native actions and approvals into the same artifact view
- connect `audit.action.logged` taxonomy work when runtime event forwarding is added
---
## 4. Commits
- `e294cf5`