You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Adds the drizzle-engineering skill source under .agents/skills-src/drizzle-engineering/ (flattened to match the sqlite-engineering layout) plus a UserPromptSubmit pre-hook at assets/hooks/skill_prehook.py, adapted from the sqlite pre-hook with Drizzle/Turso/ORM triggers.
The installed copy lives at ~/.agents/skills/drizzle-engineering/ and the hook is wired in ~/.kimi-code/config.toml alongside the sqlite hook (both user-scope, not in this repo).
Verified: hook compiles, injects the skill on a drizzle-flavored prompt, no-ops silently on an unrelated prompt, and config.toml parses with both hook entries.
CodeAnt-AI Description
Add Drizzle ORM guidance and automatic skill loading for database work
What Changed
Adds a Drizzle engineering skill covering schema design, type-safe queries, validation, transactions, performance, and migration safety across supported database drivers
Automatically provides the skill when prompts contain Drizzle-specific terms, while avoiding unrelated database prompts
Falls back to a clear instruction to read the skill when the skill file cannot be loaded, and reports hook failures without interrupting prompt processing
Documents safe migration workflows, atomic multi-row writes, protection against unsafe SQL, N+1 queries, timestamp mismatches, and destructive schema changes
Impact
✅ Consistent Drizzle guidance for ORM tasks ✅ Fewer unsafe production migrations ✅ Fewer accidental SQL injection and partial-write risks
💡 Usage Guide
Checking Your Pull Request
Every time you make a pull request, our system automatically looks through it. We check for security issues, mistakes in how you're setting up your infrastructure, and common code problems. We do this to make sure your changes are solid and won't cause any trouble later.
Talking to CodeAnt AI
Got a question or need a hand with something in your pull request? You can easily get in touch with CodeAnt AI right here. Just type the following in a comment on your pull request, and replace "Your question here" with whatever you want to ask:
@codeant-ai ask: Your question here
This lets you have a chat with CodeAnt AI about your pull request, making it easier to understand and improve your code.
Example
@codeant-ai ask: Can you suggest a safer alternative to storing this secret?
Preserve Org Learnings with CodeAnt
You can record team preferences so CodeAnt AI applies them in future reviews. Reply directly to the specific CodeAnt AI suggestion (in the same thread) and replace "Your feedback here" with your input:
@codeant-ai: Your feedback here
This helps CodeAnt AI learn and adapt to your team's coding style and standards.
Example
@codeant-ai: Do not flag unused imports.
Retrigger review
Ask CodeAnt AI to review the PR again, by typing:
@codeant-ai: review
Check Your Repository Health
To analyze the health of your code repository, visit our dashboard at https://app.codeant.ai. This tool helps you identify potential issues and areas for improvement in your codebase, ensuring your repository maintains high standards of code health.
@Bonobo791, you've reached your PR review limit, so we couldn't start this review.
Next review available in:21 minutes
Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).
How can I continue?
After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.
To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.
How do review limits work?
CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.
For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.
Added automatic detection of Drizzle-related prompts to provide relevant engineering guidance during development.
Documentation
Added a Drizzle engineering skill covering schema design, migrations, queries, transactions, validation, and database drivers.
Added references for migration workflows, query performance, schema declarations, transactions, driver behavior, and validation practices.
Documented safety checks, rollback procedures, security considerations, and post-migration verification.
Walkthrough
Adds a Drizzle ORM engineering skill with schema, query, validation, migration, transaction, driver, and troubleshooting guidance. Adds a prompt hook that detects Drizzle-related requests and injects the full skill or a directive.
This PR adds a prompt prehook that detects Drizzle-related work, loads the engineering skill, and injects its guidance into the agent context before processing continues.
The reason will be displayed to describe this comment to others. Learn more.
Review complete. The code changes add comprehensive Drizzle ORM engineering documentation and tooling that are well-structured with no blocking defects identified. All Python code includes proper error handling, and documentation examples are syntactically correct.
You can now have the agent implement changes and create commits directly on your pull request's source branch. Simply comment with /q followed by your request in natural language to ask the agent to make changes.
Bonobo791
changed the title
Phase D: team management UI + invites
chore: drizzle-engineering skill source + pre-hook
Aug 3, 2026
• Add a Drizzle ORM engineering skill with non-negotiable safety and workflow rules.
• Document Drizzle schema, migrations, performance, transactions, and drizzle-zod validation
guidance.
• Add a prompt prehook that auto-injects the skill when Drizzle-related keywords appear.
Diagram
graph TD
P(["User prompt"]) --> R["Hook runner (harness)"] --> H["skill_prehook.py"] --> D{"Drizzle keywords?"}
D -->|"no"| N["No-op"]
D -->|"yes"| S[["SKILL.md"]] --> C["Injected agent context"]
S --> M[["references/*.md"]]
subgraph Legend
direction LR
_proc["Process"] ~~~ _dec{"Decision"} ~~~ _file[["File"]]
end
Loading
High-Level Assessment
The following are alternative approaches to this PR:
1. Derive hook triggers from SKILL.md front matter
➕ Single source of truth for trigger keywords (no duplicated lists).
➕ Easier to add/remove triggers without touching code.
➖ Requires parsing YAML front matter in the hook (slightly more complexity).
➖ More failure modes if the markdown format changes.
➕ Scales to many skills with the same JSON-in/emit-text-out protocol.
➕ Centralizes error handling, env var conventions, and logging.
➖ Overhead if this repo will only ever have one or two skills.
➖ Adds indirection that can make hook behavior harder to debug.
Recommendation: The PR’s approach (a simple, fail-safe keyword prehook plus comprehensive docs) is appropriate for low-friction adoption. If more skills are expected, consider consolidating keyword definitions by reading triggers from the skill’s front matter, and/or introducing a small generic hook runner to avoid per-skill bespoke scripts.
Files changed (7) +899 / -0
Documentation (6) +791 / -0
SKILL.mdAdd Drizzle engineering skill doctrine and safety defaults+98/-0
Add Drizzle engineering skill doctrine and safety defaults
• Introduces the drizzle-engineering skill with strict operational defaults for schema-as-source-of-truth, migration workflows, expand-and-contract, transaction discipline, and raw SQL safety. Includes failure-mode guidance and verification habits, plus pointers to detailed reference docs.
migrations-workflow.mdDocument drizzle-kit migration lifecycle and drift recovery+159/-0
Document drizzle-kit migration lifecycle and drift recovery
• Provides an operational guide for drizzle-kit commands, when to generate vs push, how to review generated SQL, and how to handle custom DML migrations. Covers expand-and-contract, application methods (CLI/programmatic/D1), journal immutability, and rollback strategy.
queries-performance.mdDocument safe query shapes and Drizzle performance rules+189/-0
Document safe query shapes and Drizzle performance rules
• Adds guidance on choosing core vs relational query APIs, eliminating N+1 patterns, batching and upserts, prepared statements, pagination/counting, and strict raw SQL safety rules. Emphasizes explicit selects and observable SQL for performance debugging.
schema-declaration.mdDocument schema declaration patterns and drizzle-kit diff correctness+182/-0
Document schema declaration patterns and drizzle-kit diff correctness
• Defines conventions for drizzle.config.ts, column modes (notably SQLite timestamp units), defaults, primary keys, and constraints/indexes. Includes guidance for self-references, relations() as an N+1 defense, and multi-file schema organization.
transactions-drivers.mdDocument db.transaction semantics across drivers+100/-0
Document db.transaction semantics across drivers
• Explains how Drizzle transactions behave, the importance of using tx within callbacks, nested transactions via savepoints, and SQLite BEGIN behavior selection. Summarizes driver-specific traps (Turso/libsql, D1, Neon HTTP, Postgres) and prescribes transaction discipline.
• Adds guidance on generating insert/select schemas from the Drizzle schema, refining without forking, and enforcing boundary discipline. Highlights avoiding mass-assignment vectors by omitting server-owned fields and using strict request parsing.
skill_prehook.pyAdd prompt prehook to auto-inject Drizzle skill on keyword match+108/-0
Add prompt prehook to auto-inject Drizzle skill on keyword match
• Adds a harness-compatible prehook that reads the user prompt (JSON or raw text), matches Drizzle-related trigger patterns, and emits either the full SKILL.md or a short directive depending on env configuration. Designed to be non-blocking: always exits 0 and swallows unexpected exceptions.
The reason will be displayed to describe this comment to others. Learn more.
Pull Request Overview
The pull request is currently not up to standards according to Codacy, primarily due to issues within the pre-hook implementation and documentation accuracy. The most significant concern is the overly broad trigger list in the Python pre-hook, which will cause the Drizzle engineering skill to be injected into almost any database-related task, regardless of whether Drizzle is actually used. This will lead to context window pollution and potentially conflicting instructions.
Additionally, the documentation contains outdated information regarding drizzle-kit's ability to detect renames and recommends unsafe patterns using sql.raw. The pre-hook script itself is complex and lacks automated tests, posing a maintenance risk.
About this PR
Generic triggers such as 'orm' and 'migration' are highly likely to cause false positives, injecting Drizzle-specific guidelines into tasks that use other ORMs (e.g., Prisma, SQLAlchemy) or generic migration tools.
The pre-hook script is complex and lacks automated unit tests. Relying on manual verification for a component that modifies the agent's context is risky; consider adding unit tests to verify trigger logic and prompt parsing.
Test suggestions
Detection of Drizzle-specific keywords (e.g., 'drizzle-orm', 'sqliteTable')
Parsing of JSON input from standard agent harnesses like Claude Code
Switching between full skill inlining and directive mode via DRIZZLE_ENGINEERING_HOOK_FULL
Handling of file access errors when reading SKILL.md to ensure the hook remains silent rather than failing
Verification that broad keywords like 'orm' or 'migration' do not trigger excessive false positives for non-Drizzle tasks
Unit tests for skill_prehook.py to address high complexity and lack of coverage
Prompt proposal for missing tests
Consider implementing these tests if applicable:
1. Detection of Drizzle-specific keywords (e.g., 'drizzle-orm', 'sqliteTable')
2. Parsing of JSON input from standard agent harnesses like Claude Code
3. Switching between full skill inlining and directive mode via DRIZZLE_ENGINEERING_HOOK_FULL
4. Handling of file access errors when reading SKILL.md to ensure the hook remains silent rather than failing
5. Verification that broad keywords like 'orm' or 'migration' do not trigger excessive false positives for non-Drizzle tasks
6. Unit tests for skill_prehook.py to address high complexity and lack of coverage
The reason will be displayed to describe this comment to others. Learn more.
🔴 HIGH RISK
The trigger list includes extremely broad terms like 'orm', 'migration', 'select', 'join', and 'upsert'. These will cause the Drizzle engineering skill to be injected into the context for nearly any database-related task, polluting the context window. Narrow these to Drizzle-specific terms such as package names ('drizzle-orm'), specific table builders ('pgTable', 'sqliteTable'), or unique API methods.
The reason will be displayed to describe this comment to others. Learn more.
🟡 MEDIUM RISK
Avoid silently ignoring all exceptions. Even if the hook must not block the prompt flow, logging the error to sys.stderr ensures visibility for troubleshooting when the skill fails to load. Update the exception handling in the __main__ block to log any caught Exception to sys.stderr before exiting with code 0.
The reason will be displayed to describe this comment to others. Learn more.
🟡 MEDIUM RISK
Suggestion: Using sql.raw is discouraged even for internal identifiers. Use the sql template literal with the column object instead to ensure proper escaping.
The reason will be displayed to describe this comment to others. Learn more.
🟡 MEDIUM RISK
Suggestion: Referencing the excluded table using sql.raw is unnecessary and bypasses safety mechanisms. Drizzle can safely interpolate column identifiers within the sql template tag (e.g., sqlexcluded.${users.name}`).
The reason will be displayed to describe this comment to others. Learn more.
⚪ LOW RISK
The claim that drizzle-kit cannot see renames is outdated. The CLI provides interactive prompts to detect renames during the generate command, which results in correct rename statements in the migration file.
The reason will be displayed to describe this comment to others. Learn more.
⚪ LOW RISK
Suggestion: This file duplicates the read_prompt logic from sqlite-engineering. Consider abstracting this into a shared utility or using a template to avoid manually syncing boilerplate across every new engineering skill.
The entrypoint wraps main() in a blanket except Exception: pass and always exits 0, so
unexpected runtime failures silently disable skill injection and are difficult to diagnose. This
preserves prompt flow as intended, but provides no observability to detect/debug hook breakage.
+ try:+ main()+ except Exception:+ pass # hooks must never break the prompt flow
Relevance
●● Moderate
No historical evidence: new hook file not present in repo history; no prior guidance on logging vs
silent hooks.
ⓘ Recommendations generated based on similar findings in past PRs
Evidence
The script explicitly suppresses any exception thrown by main() and still exits 0, meaning
failures produce no output/status signal for the harness and are hard to debug.
The issue below was found during a code review. Follow the provided context and guidance below and implement a solution
### Issue description
The hook suppresses all unexpected exceptions at the entrypoint and always exits 0, which makes injection failures indistinguishable from a normal no-op. This complicates diagnosing why the skill is not being injected.
### Issue Context
The hook’s design goal is to never break the prompt flow; changes must preserve that behavior.
### Fix Focus Areas
- .agents/skills-src/drizzle-engineering/assets/hooks/skill_prehook.py[103-108]
### Implementation notes
- Keep `exit(0)` behavior.
- Add optional diagnostics gated by an env var (e.g., `DRIZZLE_ENGINEERING_HOOK_DEBUG=1`):
- On exception, write a short error message (and optionally traceback) to `stderr`.
- Narrow the catch if desired (e.g., avoid swallowing `SystemExit`), but ensure the hook still can’t break the harness.
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
The prehook trigger \bselect\s+.+\s+from\b won’t match common multiline SQL (e.g., `SELECT\n
...\nFROM ...) because .` doesn’t match newlines in Python regex by default, so some Drizzle/SQL
prompts won’t inject the skill. This reduces reliability of automatic skill loading for formatted
SQL/code-block prompts that don’t contain other Drizzle keywords.
No repo history: path '.agents/skills-src/…' absent on default branch; can’t infer trigger-regex
expectations.
ⓘ Recommendations generated based on similar findings in past PRs
Evidence
The trigger list includes r"\bselect\s+.+\s+from\b" and the regex is compiled without re.DOTALL,
so . cannot cross newlines and multiline SQL won’t satisfy this trigger.
The issue below was found during a code review. Follow the provided context and guidance below and implement a solution
### Issue description
The prompt trigger regex uses `\bselect\s+.+\s+from\b`, where `.` does not match newline characters in Python regex by default. This causes false negatives for multiline SQL snippets.
### Issue Context
This hook is meant to inject the drizzle-engineering skill whenever prompts touch Drizzle/SQL work, including code blocks that often format SQL across multiple lines.
### Fix Focus Areas
- .agents/skills-src/drizzle-engineering/assets/hooks/skill_prehook.py[28-44]
### Implementation notes
- Replace the `select ... from` trigger with a newline-tolerant, non-greedy variant, e.g.:
- `r"\bselect\s+[\s\S]+?\s+from\b"`
- (Optional) Add a reasonable upper bound to avoid scanning extremely long prompts for this pattern (e.g., via a more constrained pattern), while keeping other triggers unchanged.
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.agents/skills-src/drizzle-engineering/assets/hooks/skill_prehook.py:
- Around line 103-108: Update the __main__ exception handling so hook failures
are no longer silently discarded: catch expected decoding and I/O failures at
their sources, and if the final catch-all remains necessary to preserve exit
code 0, write only the exception type to stderr without including prompt or
skill contents.
- Around line 2-4: Update the module docstring at the top of skill_prehook.py so
the opening line is followed by a blank line, the summary appears on the next
paragraph’s first line, and that summary ends with a period, preserving the
existing description.
- Around line 35-40: In the trigger patterns used by the prehook, narrow generic
matches such as join, relations, and migrations so unrelated prompts do not
enable full SKILL.md injection. Require a Drizzle-specific term or database
context for these generic terms, while preserving detection of valid Drizzle
prompts. Add regression coverage for unrelated prompts and valid Drizzle prompts
through the existing hook test symbols.
In @.agents/skills-src/drizzle-engineering/references/queries-performance.md:
- Around line 84-91: Update the onConflictDoUpdate example to remove the
undefined excludedName reference in set; replace it with a complete valid
SQL/value expression consistent with the insert example, or omit the example if
no valid expression is intended.
- Around line 55-60: In the N+1 query example, update the result binding in the
`posts` query from `posts` to `postRows` so it does not shadow the table
reference before initialization. Preserve the loop and author assignment logic,
updating any references to the fetched result as needed.
- Around line 96-102: Replace the blanket buildConflictUpdateColumns example
with a valid Drizzle pattern: define a local helper or explicit set map that
includes only mutable user column keys, excluding immutable/defaulted fields
such as id and createdAt, and remove the invalid buildConflictUpdateColumns
import and usage.
In @.agents/skills-src/drizzle-engineering/references/schema-declaration.md:
- Around line 24-30: Update the defineConfig example to clearly label it as
SQLite-only, or align it with the repository’s Turso setup by using dialect
'turso' and the corresponding `@libsql/client` credentials including the auth
token. Ensure the schema, output, and strict/verbose settings remain unchanged.
- Around line 124-135: Update the self-reference guidance in the schema
declaration documentation to reflect that drizzle-orm@0.45.2 supports the inline
references form when the callback explicitly returns AnySQLiteColumn. Retain or
additionally show the standalone foreignKey alternative, and ensure the example
and surrounding explanation no longer claim inline self-references are
unsupported.
In @.agents/skills-src/drizzle-engineering/SKILL.md:
- Around line 33-34: Revise the source-of-truth guidance in the TypeScript
schema section to apply only to normal schema changes, allowing manual database
reconciliation when recovering from a partial migration. Reference the
documented recovery procedure in migrations-workflow.md, while preserving the
warning against routine out-of-band edits.
- Around line 58-61: Add shell language identifiers to the command fences at
.agents/skills-src/drizzle-engineering/SKILL.md lines 58-61 and
.agents/skills-src/drizzle-engineering/references/migrations-workflow.md lines
62-64, 95-97, and 119-123; no other content changes are needed.
- Around line 46-47: Update the multi-row mutation rule in the Drizzle
engineering guidance to require db.transaction when the driver supports it,
while allowing an equivalent atomic SQL statement or driver-supported batch API
such as db.batch when transactions are unavailable. Preserve the requirement
that the entire mutation remains atomic and cannot leave a partial write.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
Push a commit to this branch (recommended)
Create a new PR with the fixes
ℹ️ Review info⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: bfe8d7e3-1229-47ff-bb34-5ea23120fb94
📥 Commits
Reviewing files that changed from the base of the PR and between cd3283a and adc7a21.
The reason will be displayed to describe this comment to others. Learn more.
🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
Align or label the database configuration example.
The repository configuration in drizzle.config.ts, Lines 31-41, uses dialect: 'turso', @libsql/client, and an auth token. This example sets dialect: 'sqlite' and provides only a URL. If contributors copy it into this repository, Drizzle Kit can use the wrong dialect and driver configuration. Mark this as a SQLite-only example or show the repository's Turso configuration.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In @.agents/skills-src/drizzle-engineering/references/schema-declaration.md
around lines 24 - 30, Update the defineConfig example to clearly label it as
SQLite-only, or align it with the repository’s Turso setup by using dialect
'turso' and the corresponding `@libsql/client` credentials including the auth
token. Ensure the schema, output, and strict/verbose settings remain unchanged.
skill_prehook.py silent exception (Codacy/Qodo/CodeRabbit): catch-all now writes the exception type and message to stderr; still exits 0 so the prompt flow never breaks. No prompt/skill content is logged.
skill_prehook.py docstring D205/D213/D415 (CodeRabbit trivial): summary moved to its own line, ends with a period.
queries-performance.md N+1 example (CodeRabbit Minor): const posts = await db.select().from(posts) was a TDZ ReferenceError — result renamed to postRows.
queries-performance.md upsert example (CodeRabbit Minor + Major): undefined excludedName placeholder replaced with the complete sql.raw expression; buildConflictUpdateColumns is NOT a drizzle-orm export (verified against 0.45.2) — example now defines the helper locally per the Drizzle upsert guide and passes explicit mutable column keys (no id/createdAt).
schema-declaration.md self-reference section (CodeRabbit Major): prose claimed inline self-references are unsupported while the example showed the typed inline form — prose corrected: inline works with an explicit AnySQLiteColumn return type; standalone foreignKey noted as the circular-reference solution.
schema-declaration.md config example (CodeRabbit Major): labeled as a generic SQLite example with a pointer to this repo's dialect: 'turso' + @libsql/client setup in drizzle.config.ts.
SKILL.md rule 1 (CodeRabbit Major): qualified with the documented exception — manual reconciliation after a partially applied migration, per migrations-workflow.md.
SKILL.md rule 7 (CodeRabbit Major): now allows a single atomic statement or db.batch on drivers that reject SQL transactions (D1); atomicity requirement unchanged.
migrations-workflow.md rename claim (Codacy LOW): softened — with strict: true kit prompts interactively; non-interactive runs (CI, agents) still get drop+add.
MD040 fences (CodeRabbit trivial): text/sh identifiers added to the bare fences in SKILL.md and migrations-workflow.md.
Skipped (invalid or moot):
Codacy MEDIUM ×2 — sql.raw for the excluded. pseudo-table: this is the one use the doc's own rule 1 explicitly carves out as legitimate (identifiers are hardcoded from your own schema, and the quoting form excluded."col" is the documented pattern). No change.
Qodo — select\s+.+\s+from doesn't match newlines: moot, that trigger was removed entirely as too broad.
Codacy LOW — read_prompt duplication across skill hooks: the hooks are independent single-file installs wired per-skill; a shared module would couple separate install trees. Matches the sqlite-engineering precedent. No change.
Hook re-verified after changes: compiles, injects on Drizzle prompts, silent no-op (exit 0) on generic database and unrelated prompts. Installed copy at ~/.agents/skills/drizzle-engineering/ synced identical to this tree.
Overlapping Drizzle triggers cause duplicate full skill injection when both hooks are enabled
When both the documented sqlite and Drizzle hooks are installed, this drizzle trigger overlaps with the sqlite hook's own drizzle trigger. In the default full mode, every Drizzle prompt causes both complete skill documents and directives to be appended, unnecessarily consuming context and potentially crowding out the user request. Avoid overlapping triggers or coordinate the hooks so only one full payload is emitted.
⚠️ Every Drizzle prompt receives duplicate skill payloads.
⚠️ Duplicate context consumes available agent context capacity.
(Use Cmd/Ctrl + Click for best experience)
Prompt for AI Agent 🤖
This is a comment left during a code review.
**Path:** .agents/skills-src/drizzle-engineering/assets/hooks/skill_prehook.py
**Line:** 35:36
**Comment:***Performance: When both the documented sqlite and Drizzle hooks are installed, this `drizzle` trigger overlaps with the sqlite hook's own `drizzle` trigger. In the default full mode, every Drizzle prompt causes both complete skill documents and directives to be appended, unnecessarily consuming context and potentially crowding out the user request. Avoid overlapping triggers or coordinate the hooks so only one full payload is emitted.
Validate the correctness of the flagged issue. If correct, How can I resolve this? If you propose a fix, implement it and please make it concise.
Once fix is implemented, also check other comments on the same PR, and ask user if the user wants to fix the rest of the comments as well. if said yes, then fetch all the comments validate the correctness and implement a minimal fix
Major
2026-08-03 12:33
Api mismatch
Unqualified transaction guidance can generate unsupported code for D1 drivers
This directive unconditionally instructs the agent to put multi-row mutations inside db.transaction, but the loaded skill explicitly says D1 rejects SQL transactions and requires a single atomic statement or db.batch instead. For D1/Turso work, the injected directive can cause unsupported transaction code. Qualify this instruction by driver.
⚠️ D1 implementations require manual correction to use db.batch.
(Use Cmd/Ctrl + Click for best experience)
Prompt for AI Agent 🤖
This is a comment left during a code review.
**Path:** .agents/skills-src/drizzle-engineering/assets/hooks/skill_prehook.py
**Line:** 53:54
**Comment:***Api Mismatch: This directive unconditionally instructs the agent to put multi-row mutations inside `db.transaction`, but the loaded skill explicitly says D1 rejects SQL transactions and requires a single atomic statement or `db.batch` instead. For D1/Turso work, the injected directive can cause unsupported transaction code. Qualify this instruction by driver.
Validate the correctness of the flagged issue. If correct, How can I resolve this? If you propose a fix, implement it and please make it concise.
Once fix is implemented, also check other comments on the same PR, and ask user if the user wants to fix the rest of the comments as well. if said yes, then fetch all the comments validate the correctness and implement a minimal fix
Round-2 dispositions at fa7fcf8 (CodeAnt suggestions on 926697e) — both verified and fixed:
Trigger overlap (Major): valid. The sqlite-engineering hook had drizzle/drizzle-kit triggers predating this hook, so every Drizzle prompt injected BOTH full skills. Removed those two triggers from the sqlite hook (with a comment explaining why) — the drizzle hook now owns the ORM layer, the sqlite hook the engine layer. Verified: a drizzle-only prompt now fires only the drizzle hook; an engine-only prompt only the sqlite hook. Residual dual injection happens only when a prompt names both layers (e.g. "drizzle migration on turso"), which is the pairing both skills explicitly prescribe.
Unqualified db.transaction in the DIRECTIVE (Major): valid — the directive contradicted the amended SKILL.md rule 7. Directive now reads "multi-row mutations stay atomic (db.transaction where the driver supports it; on drivers that reject SQL transactions like D1, a single atomic statement or db.batch)".
Codacy's remaining "1 minor (Documentation)" is the previously-dispositioned read_prompt duplication note — no change.
Both installed hooks at ~/.agents/skills/ synced identical; both compile and smoke-tested.
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.agents/skills-src/drizzle-engineering/references/migrations-workflow.md:
- Around line 120-123: Update the migration workflow around the drizzle-kit
generate and wrangler d1 migrations apply commands so both tools use the same
migrations directory. Configure Wrangler’s migrations_dir to drizzle, configure
Drizzle to emit to Wrangler’s configured directory, or add a documented copy
step before applying migrations; ensure the documented commands operate on the
generated files.
In @.agents/skills-src/drizzle-engineering/references/queries-performance.md:
- Around line 94-113: Update the upsert example around
buildConflictUpdateColumns so the set property is passed inside an
onConflictDoUpdate({ ... }) options object, keeping the helper as the value for
the set field and removing the standalone set expression.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
Push a commit to this branch (recommended)
Create a new PR with the fixes
ℹ️ Review info⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 2047b1ca-5cfa-472e-998c-57c56def4fc6
📥 Commits
Reviewing files that changed from the base of the PR and between adc7a21 and 926697e.
Cloudflare Wrangler D1 migrations apply migrations_dir default official documentation
💡 Result:
In Cloudflare Wrangler, the default directory for D1 migrations is./migrations [1][2]. When you run wrangler d1 migrations create, Wrangler automatically creates this migrations folder in your project directory if it does not already exist [3][4]. You can customize this location in your wrangler.jsonc or wrangler.toml configuration file by specifying the migrations_dir field within your D1 database binding configuration [3][5][6]. Example configuration for a custom directory: { "d1_databases": [ { "binding": "DB", "database_name": "my-database", "database_id": "", "migrations_dir": "my-custom-migrations-folder" } ] } If you are using an ORM that requires a nested directory structure (such as Drizzle), you must set both migrations_dir and migrations_pattern to correctly locate and apply your migration files [3][5][7]. The migrations_pattern defaults to <migrations_dir>/*.sql if not otherwise specified [5][6][7].
Configure Wrangler to read Drizzle's output directory.
drizzle-kit generate writes to ./drizzle, while Wrangler defaults to ./migrations when migrations_dir is unset. Set Wrangler's migrations_dir to drizzle, configure Drizzle to use Wrangler's directory, or document a copy step.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In @.agents/skills-src/drizzle-engineering/references/migrations-workflow.md
around lines 120 - 123, Update the migration workflow around the drizzle-kit
generate and wrangler d1 migrations apply commands so both tools use the same
migrations directory. Configure Wrangler’s migrations_dir to drizzle, configure
Drizzle to emit to Wrangler’s configured directory, or add a documented copy
step before applying migrations; ensure the documented commands operate on the
generated files.
The reason will be displayed to describe this comment to others. Learn more.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bashset -e
file=".agents/skills-src/drizzle-engineering/references/queries-performance.md"
sed -n '70,125p'"$file"printf'\n--- nearby onConflictDoUpdate occurrences ---\n'
rg -n -C 4 'onConflictDoUpdate|buildConflictUpdateColumns|^set:'"$file"printf'\n--- repository references to the helper ---\n'
rg -n -C 3 'buildConflictUpdateColumns'.
Repository: Bonobo791/Moderaty
Length of output: 3373
Wrap the helper call in onConflictDoUpdate({ ... }). The standalone set: at Line 113 is invalid TypeScript after the preceding call closes.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In @.agents/skills-src/drizzle-engineering/references/queries-performance.md
around lines 94 - 113, Update the upsert example around
buildConflictUpdateColumns so the set property is passed inside an
onConflictDoUpdate({ ... }) options object, keeping the helper as the value for
the set field and removing the standalone set expression.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
codeant-ai
Bot
added
the
size:XL
![amazon-q-developer[bot]](https://avatars.githubusercontent.com/in/1220912?s=60&v=4){kind=small}
Bonobo791
changed the title
![codacy-production[bot]](https://avatars.githubusercontent.com/in/56611?s=60&v=4){kind=small}
![qodo-code-review[bot]](https://avatars.githubusercontent.com/in/484649?s=60&v=4){kind=small}
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
User description
Adds the drizzle-engineering skill source under
.agents/skills-src/drizzle-engineering/(flattened to match the sqlite-engineering layout) plus a UserPromptSubmit pre-hook atassets/hooks/skill_prehook.py, adapted from the sqlite pre-hook with Drizzle/Turso/ORM triggers.The installed copy lives at
~/.agents/skills/drizzle-engineering/and the hook is wired in~/.kimi-code/config.tomlalongside the sqlite hook (both user-scope, not in this repo).Verified: hook compiles, injects the skill on a drizzle-flavored prompt, no-ops silently on an unrelated prompt, and config.toml parses with both hook entries.
CodeAnt-AI Description
Add Drizzle ORM guidance and automatic skill loading for database work
What Changed
Impact
✅ Consistent Drizzle guidance for ORM tasks✅ Fewer unsafe production migrations✅ Fewer accidental SQL injection and partial-write risks💡 Usage Guide
Checking Your Pull Request
Every time you make a pull request, our system automatically looks through it. We check for security issues, mistakes in how you're setting up your infrastructure, and common code problems. We do this to make sure your changes are solid and won't cause any trouble later.
Talking to CodeAnt AI
Got a question or need a hand with something in your pull request? You can easily get in touch with CodeAnt AI right here. Just type the following in a comment on your pull request, and replace "Your question here" with whatever you want to ask:
This lets you have a chat with CodeAnt AI about your pull request, making it easier to understand and improve your code.
Example
Preserve Org Learnings with CodeAnt
You can record team preferences so CodeAnt AI applies them in future reviews. Reply directly to the specific CodeAnt AI suggestion (in the same thread) and replace "Your feedback here" with your input:
This helps CodeAnt AI learn and adapt to your team's coding style and standards.
Example
Retrigger review
Ask CodeAnt AI to review the PR again, by typing:
Check Your Repository Health
To analyze the health of your code repository, visit our dashboard at https://app.codeant.ai. This tool helps you identify potential issues and areas for improvement in your codebase, ensuring your repository maintains high standards of code health.