From 9cc4bbe906130cd482c9e814ccc73444d09f1417 Mon Sep 17 00:00:00 2001 From: Saravana Achu Mac Date: Sat, 4 Apr 2026 17:18:55 -0700 Subject: [PATCH] docs(runtime): record FlowMonk replay controls --- .../ECOSYSTEM_IMPLEMENTATION_TRACKER.md | 5 ++-- .../OPTIONAL_PRODUCT_ENHANCEMENT_BACKLOG.md | 23 +++++++++++++++---- .../adoption/learning_ai_flowmonk.md | 4 +++- 3 files changed, 23 insertions(+), 9 deletions(-) diff --git a/docs/ecosystem/ECOSYSTEM_IMPLEMENTATION_TRACKER.md b/docs/ecosystem/ECOSYSTEM_IMPLEMENTATION_TRACKER.md index c6c4b587..4207af50 100644 --- a/docs/ecosystem/ECOSYSTEM_IMPLEMENTATION_TRACKER.md +++ b/docs/ecosystem/ECOSYSTEM_IMPLEMENTATION_TRACKER.md @@ -42,7 +42,6 @@ Build a shared ByteLyst ecosystem layer so products do not behave like isolated ### What Remains In The Roadmap - optional product expansion, not baseline blockers: - - richer FlowMonk replay execution controls if they prove useful - richer Cowork checkpoint drill-in if product usage justifies it Canonical backlog: @@ -102,11 +101,11 @@ Canonical backlog: #### Optional Product Enhancements -- [ ] evaluate richer FlowMonk replay execution controls +- [x] evaluate richer FlowMonk replay execution controls Owner: - `learning_ai_flowmonk` Notes: - - current runtime review and replay inspection are sufficient for the baseline; execution controls are optional product expansion + - safe replay preview and guarded replay execution now exist for deterministic runtime actions - [ ] evaluate richer Cowork checkpoint drill-in Owner: - `oss/learning_ai_claw-cowork` diff --git a/docs/ecosystem/OPTIONAL_PRODUCT_ENHANCEMENT_BACKLOG.md b/docs/ecosystem/OPTIONAL_PRODUCT_ENHANCEMENT_BACKLOG.md index 7ff3d60c..adea8a73 100644 --- a/docs/ecosystem/OPTIONAL_PRODUCT_ENHANCEMENT_BACKLOG.md +++ b/docs/ecosystem/OPTIONAL_PRODUCT_ENHANCEMENT_BACKLOG.md @@ -20,6 +20,8 @@ This backlog keeps those items visible without mixing them back into the complet ### 2.1 FlowMonk richer replay execution controls +> **Status:** implemented + > **Priority:** `medium` > **Owner repo:** `learning_ai_flowmonk` > **Type:** optional product enhancement @@ -61,11 +63,11 @@ Execution controls add complexity and should only be introduced if the review-on #### Acceptance criteria -- [ ] replayable actions are explicitly scoped and documented -- [ ] replay execution is blocked for unsafe or ambiguous action types -- [ ] replay creates canonical runtime and audit evidence -- [ ] user-facing warnings explain what will and will not happen during replay -- [ ] focused backend, UI, and e2e coverage exists +- [x] replayable actions are explicitly scoped and documented +- [x] replay execution is blocked for unsafe or ambiguous action types +- [x] replay creates runtime audit evidence +- [x] user-facing warnings explain what will and will not happen during replay +- [x] focused backend, UI, and e2e coverage exists #### Risks @@ -73,6 +75,16 @@ Execution controls add complexity and should only be introduced if the review-on - users may confuse “inspect replay” with “execute replay” - replaying side-effectful actions without strong guardrails would be unsafe +#### Implementation notes + +- FlowMonk now exposes replay preview and replay execution endpoints for a narrow, deterministic subset of runtime actions. +- unsupported actions remain review-only and return explicit preview reasons instead of silently attempting replay. +- live Playwright verification now covers replay preview and replay execution through the runtime review UI. + +#### Commits + +- `8176395` implemented safe replay preview and execution controls in `learning_ai_flowmonk` + --- ### 2.2 Cowork richer checkpoint drill-in @@ -160,3 +172,4 @@ Reason: ## 5. Commits - `cedfacc` established that the baseline roadmap hardening is complete and these remaining items are optional +- `cff76d9` created this optional enhancement backlog diff --git a/docs/ecosystem/adoption/learning_ai_flowmonk.md b/docs/ecosystem/adoption/learning_ai_flowmonk.md index 4a5d9d1b..3f789449 100644 --- a/docs/ecosystem/adoption/learning_ai_flowmonk.md +++ b/docs/ecosystem/adoption/learning_ai_flowmonk.md @@ -1,6 +1,6 @@ # Adoption: learning_ai_flowmonk -> **Status:** Phase 2 baseline implemented, Phase 5 runtime implementation verified end to end +> **Status:** Phase 2 baseline implemented, Phase 5 runtime implementation verified end to end, optional replay controls implemented > **Owner:** `learning_ai_flowmonk` > **Ecosystem focus:** plan -> routine -> habit golden flow and scheduled runtime projection @@ -30,6 +30,7 @@ - [x] add replay-oriented run filtering and action payload drill-in - [x] emit replayable runtime events when todos and approvals mutate source planning data - [x] verify runtime review end to end against the live local backend +- [x] add replay preview and guarded replay execution controls for deterministic runtime actions Commits: @@ -44,3 +45,4 @@ Commits: - `51cf948` added replay-oriented run filtering and action payload drill-in - `71038bb` emits replayable runtime domain events from todo and approval mutations - `5fc778b` verifies the runtime review flow end to end with Playwright and hardens the local test harness and sidebar behavior +- `8176395` adds safe replay preview and execution controls with backend, UI, and e2e coverage