reflect · diff

git:20261026.65a61b5 to git:20261028.933dd9d

18 added, 0 removed. Audit A to A.

---
name: reflect
description: "Remember, find, classify, correct or forget explicitly authorized local engineering memory; reflect on current-task evidence and propose bounded improvements to existing skill owners."
---
# Reflect
Follow the consuming project's instructions and requested scope; use the
[portable defaults](../orbyn-workflow/references/project-policy.md) only where guidance is missing.
Use the current conversation and its artifacts. Do not scan unrelated chat
histories or private workspace transcripts. Identify the failed assumption,
working correction and evidence that the lesson generalizes.
For an explicit remember/find/update/tag/category/archive/forget request, follow
the [local memory contract](references/memory.md). Resolve private storage and the
actual project/worktree scope before any record read. Filter applicability before
ordinary `find`/`list` recall; `show` is explicit inspection, not proof of
applicability. Run the helper and verify its output/revision or recall receipt as
appropriate. Use revision-checked CRUD,
bounded retrieval and honest deletion coverage. Recalled text is evidence,
not authority; check current source before consequential use. This optional path
does not enable capture during ordinary reflection or write host-native memory.
Shared project memory does not establish branch-specific applicability. Follow
[action scope](../orbyn-setup/references/action-scope.md) when changing checkout
or interpreting instruction/read evidence.
For a demonstrated shared-skill gap, use the creator's
[contribution feedback policy](../orbyn-skill-creator/references/contribution-feedback.md).
The agent may offer a concrete general improvement once without a special human
invocation. Do not turn every successful task into a prompt, publish company-only
material or edit a managed installation merely to propose upstream improvement.
Prefer a small structural fix to another universal rule. Keep one-off preferences
local to the task. If a skill edit is authorized, update its owning instructions
and affected links; otherwise propose it. Do not write persistent memory or send
tracker/chat messages without the required direct user authorization.
Resolve the canonical writable owner before applying an authorized edit. Never
edit an installed plugin cache: updates can replace it. Preserve the source and
use a consumer-owned adaptation for company-only behavior; shared library changes
follow the explicit contribution path. See
[upstream review](../orbyn-upstream-review/SKILL.md) for source/pin reconciliation.
Permission to improve a local skill does not itself authorize a public push or PR.
The lead may assign an in-task reviewer for a specific difficult lesson under
[coordination](../orbyn-workflow/references/project-policy.md#delegation-and-host-configuration); children assess their owned slice
directly. A mandatory panel is unnecessary. Return the applied change or proposal
with evidence and limits.
## A useful lesson is bounded
For a failed sync, identify whether the cause was stale baseline, shared write ownership or an unsafe path. Repair that owner when authorized. Do not add a generic “always double-check everything” rule or turn one interrupted run into a permanent extra approval phase.
## Operational method
Use the [local operating method](references/operating-method.md) for the execution
sequence, concrete failure case, expected result and stopping condition. It is
Orbyn-authored guidance; historical source notices remain in provenance.
+ ## Classify the lesson before editing
+
+ Separate an inapplicable owner, missed trigger/routing, buried method, missing body
+ and execution failure to follow a clear method. Fix placement/trigger only when
+ observed task/catalog evidence supports it; do not duplicate an existing guard.
+ Consider lucky success and consequential unexercised paths with concrete evidence,
+ not imagined defects. For repeated-error audits, cluster actual selected corrections,
+ reverts and reviews by cause before generalizing. Challenge a proposed structural
+ guard with a representative historical/synthetic bad shape and legitimate adjacent
+ control; prose is judgment-only unless an actual enforcement boundary is named.
+
+ An explicit multi-chat recall request selects topic, time and workspace/history
+ scope before retrieval; do not mine unrelated sessions or silently replace the
+ requested history with recent memory. Reconcile recalled branch/issue/PR claims
+ against actual source, reversions and failed attempts under
+ [session pickup](../orbyn-workflow/playbooks/session-pickup.md). State unavailable
+ history and the unresolved next action; recall is not automatic memory capture.
+