chore: add agent skills and tooling configs#3316
Conversation
Add an agent skill that walks through implementing a new Redis command in node-redis end-to-end: the <NAME>.ts Command file (parseCommand + transformReply), registration with JSDoc in the package commands/index.ts (raw name + camelCase alias), and a co-located <NAME>.spec.ts using parseArgs and testUtils.testAll. Includes an openai.yaml interface descriptor matching the sibling skills. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the shared .agents/ skill set (docs-sync, maintainer-review, pr-draft-summary, runtime-behavior-probe, test-coverage-improver) with their references and templates, the AGENTS.md/CLAUDE.md/GEMINI.md guidance files, and the .claude/.gemini symlinks pointing into .agents/. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
nkaradzhov
changed the title
Add a Step 0 that has the agent request the redis/redis JSON command spec (src/commands/<name>.json, or pasted from an unreleased branch) and a running Redis instance that actually has the command, since new commands are often implemented before public release. Include a spec-field -> Command mapping table (command_flags, key_specs, argument types/tokens, reply_schema, arity) and a step to verify the real reply against reply_schema. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
| -- | ||
| name: test-coverage-improver | ||
| description: 'Improve test coverage in the node-redis monorepo: run a package test suite under nyc, inspect coverage artifacts, identify low-coverage files and branches, propose high-impact tests, and confirm with the user before writing tests.' | ||
| --- |
There was a problem hiding this comment.
Invalid skill frontmatter, change line 1 to ---
| @@ -0,0 +1 @@ | |||
| ../.agents/agents No newline at end of file | |||
There was a problem hiding this comment.
This seems to be an invalid directory
| @@ -0,0 +1 @@ | |||
| ../.agents/agents No newline at end of file | |||
There was a problem hiding this comment.
This seems to be an invalid directory
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 2 potential issues.
Reviewed by Cursor Bugbot for commit 4eb4cbf. Configure here.
| @@ -0,0 +1 @@ | |||
| ../.agents/agents No newline at end of file | |||
There was a problem hiding this comment.
Agent symlinks target missing path
Medium Severity
New .claude/agents and .gemini/agents symlinks point at ../.agents/agents, but the commit only adds .agents/skills/ and never creates .agents/agents. Checkouts following these links get a broken path, so Claude and Gemini agent discovery from those entries will not work.
Additional Locations (1)
Reviewed by Cursor Bugbot for commit 4eb4cbf. Configure here.
| @@ -0,0 +1 @@ | |||
| ../.agents/agents No newline at end of file | |||
There was a problem hiding this comment.
Missing Gemini skills symlink
Low Severity
The PR wires .claude/skills to shared .agents/skills/ but does not add an equivalent .gemini/skills symlink. Gemini runners may not see the same skill set from one source, contrary to the stated goal of shared agent tooling across runners.
Additional Locations (1)
Reviewed by Cursor Bugbot for commit 4eb4cbf. Configure here.
2 participants
Description
This pull request adds the repository's shared agent tooling so Claude Code, OpenAI/Codex, and Gemini runners all work from one source: a set of skills under
.agents/skills/, theAGENTS.md/CLAUDE.md/GEMINI.mdguidance files, and.claude/.geminisymlinks pointing into.agents/. Each skill ships aSKILL.md, anagents/openai.yamlinterface descriptor for cross-runner parity, and any supportingreferences//templates/.Skills included:
<NAME>.tsCommand file, JSDoc registration incommands/index.ts, and a co-located<NAME>.spec.ts, with the monorepo conventions baked in.docs/,README.md, and per-package READMEs against the master implementation and propose targeted updates.No runtime code, tests, or build configuration changes — repository tooling only, with no behavior or backward-compatibility impact.
Checklist
npm testpass with this change (including linting)? — N/A, no runtime/build changesNote
Low Risk
Documentation and agent-config only; no published API, client behavior, or test/build pipeline changes.
Overview
This pull request adds shared agent tooling so multiple AI runners can use the same workflows from one tree under
.agents/.It introduces six skills (
docs-sync,implement-command,maintainer-review,pr-draft-summary,runtime-behavior-probe,test-coverage-improver), each with aSKILL.md, anagents/openai.yamlinterface descriptor, and supporting reference/template files where needed.AGENTS.mddocuments monorepo layout, the command pattern, and test/build conventions;CLAUDE.mdandGEMINI.mddelegate to that file..claude/and.gemini/symlinks point at.agents/for skills (and agents) so Claude and Gemini pick up the same definitions.There are no changes under
packages/, tests, or build config—repository metadata and contributor/agent guidance only.Reviewed by Cursor Bugbot for commit 4eb4cbf. Bugbot is set up for automated code reviews on this repo. Configure here.