This is an automated email from the ASF dual-hosted git repository.
zhangliang pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/shardingsphere.git
The following commit(s) were added to refs/heads/master by this push:
new 9296c26930f Add PLANS.md for codex (#37955)
9296c26930f is described below
commit 9296c26930fdf273077b26bc630429bded670ab8
Author: Liang Zhang <[email protected]>
AuthorDate: Wed Feb 4 16:07:28 2026 +0800
Add PLANS.md for codex (#37955)
---
AGENTS.md | 1 +
PLANS.md | 30 ++++++++++++++++++++++++++++++
2 files changed, 31 insertions(+)
diff --git a/AGENTS.md b/AGENTS.md
index c9e3f481742..9efb7d9555b 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -86,6 +86,7 @@ This guide is written **for AI coding agents only**. Follow
it literally; improv
## Compliance Guardrails & Checklists
- **Pre-task checklist (do before planning/coding):** re-read AGENTS.md and
`CODE_OF_CONDUCT.md`; restate user goal, constraints, forbidden tools/APIs,
coverage expectations, sandbox/network/approval limits; prefer
`rg`/`./mvnw`/`apply_patch`; avoid destructive commands (`git reset --hard`,
`git checkout --`, bulk deletes) and generated paths like `target/`.
+- **Planning reference:** keep session plans, decision logs, and
retrospectives in [PLANS.md](PLANS.md); update it when codex-readiness runs or
task scopes change.
- **Risk gate:** if any action fits the Dangerous Operation Checklist, pause
and use the confirmation template before proceeding.
- **Planning rules:** use Sequential Thinking with 3-10 actionable steps (no
single-step plans) via the plan tool for non-trivial tasks; convert all hard
requirements (SPI usage, mocking rules, coverage/test naming, forbidden APIs)
into a checklist inside the plan and do not code until each item is addressed
or explicitly waived.
- **Execution discipline:** inspect existing code before edits; keep changes
minimal; default to mocks and SPI loaders; keep variable declarations near
first use and mark retained values `final`; delete dead code and avoid
placeholders/TODOs.
diff --git a/PLANS.md b/PLANS.md
new file mode 100644
index 00000000000..f407bc7759a
--- /dev/null
+++ b/PLANS.md
@@ -0,0 +1,30 @@
+# ShardingSphere Planning Notes
+
+Use this file to keep planning context, progress checkpoints, and decision
history for work in this repository.
+
+## Purpose / Big Picture
+- Keep quality-first changes aligned with AGENTS.md and CODE_OF_CONDUCT.md
while minimizing blast radius.
+- Anchor work to ShardingSphere’s layers
(infra/database/parser/kernel/mode/jdbc/proxy/features/agent) and prefer scoped
module changes.
+- Make every plan actionable: commands, expected outputs, and verification
steps must be explicit and copy-pasteable.
+
+## Progress
+1. Confirm scope, constraints, and sandbox/approval settings; restate the user
goal in plain terms.
+2. Inspect relevant modules and tests before coding; list branches/edges that
need coverage.
+3. Implement the smallest safe change, deleting dead code and keeping
variables near first use.
+4. Verify with targeted Maven commands (e.g., `./mvnw test -pl <module>[-am]`
or `./mvnw checkstyle:check -Pcheck`) and capture exit codes.
+5. Summarize intent, edits, verification, and risks; propose next actions only
when they are natural follow-ups.
+
+## Decision Log
+- Established PLANS.md as the canonical planning reference linked from
AGENTS.md for Codex readiness checks.
+- Default workflow: run codex-readiness in read-only first; switch to execute
mode with a reviewed plan.json when command execution is required.
+- Keep planning notes in this file rather than scattering across threads to
satisfy traceability expectations.
+
+## Outcomes & Retrospective
+- Record each session’s outputs (commands + exit codes, test scopes, notable
fixes) and any remaining risks or TODOs.
+- Note when coverage, Spotless, or Checkstyle were skipped and which commands
to run later.
+- Capture rollback considerations (e.g., config toggles or revert paths) when
changes touch runtime behavior.
+
+## Surprises & Discoveries
+- Codex readiness requires a plan-named markdown or planning headings; keep
this file updated to avoid failing deterministic checks.
+- Read-only mode leaves execution checks NOT_RUN; execute mode needs a
confirmed plan.json and should be limited to scoped commands.
+- AGENTS.md line limit (<=300) and formatting expectations apply here too;
avoid placeholders and keep entries concise.
