feat: add agent skills for package registry queries and AI tool integration#39
Merged
feat: add agent skills for package registry queries and AI tool integration#39
Conversation
Two skills following the agentskills.io spec: - regxa: CLI and library API for querying package registries - regxa-ai-tool: Vercel AI SDK integration with packageTool
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (2)
📜 Recent review details⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
🧰 Additional context used🧠 Learnings (17)📚 Learning: 2026-03-10T07:36:38.679ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
📚 Learning: 2026-03-10T07:36:22.301ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
📚 Learning: 2026-03-10T07:36:54.862ZApplied to files:
📚 Learning: 2026-03-10T07:36:22.301ZApplied to files:
📚 Learning: 2026-03-10T07:36:22.301ZApplied to files:
📚 Learning: 2026-03-10T07:36:03.586ZApplied to files:
📚 Learning: 2026-03-10T07:36:46.164ZApplied to files:
📚 Learning: 2026-03-10T07:36:46.164ZApplied to files:
📚 Learning: 2026-03-10T07:36:46.164ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
📚 Learning: 2026-03-10T07:36:54.862ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
📚 Learning: 2026-03-10T07:36:29.354ZApplied to files:
🔇 Additional comments (6)
📝 WalkthroughWalkthroughAdds documentation and agent manifests for the regxa package-registry client and its Vercel AI SDK integration: SKILL docs, PURL guides and cheatsheet, API reference defining package/version/dependency/maintainer shapes and helpers, plus two agent YAML manifests for regxa and the regxa AI tool. Changes
Sequence Diagram(s)(omitted — changes are documentation and agent manifests; no new multi-component control flow requiring a sequence diagram) Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 inconclusive)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
✨ Simplify code
📝 Coding Plan
Comment |
Sequence DiagramThis PR adds two agent skills that guide how package registry queries are performed through regxa, either via direct regxa usage or through AI SDK tool integration. The flow emphasizes progressive disclosure, where detailed references are loaded only when needed before executing normalized registry operations. sequenceDiagram
participant Agent
participant SkillsRuntime
participant RegxaSkill
participant AIToolSkill
participant Regxa
participant Registry
Agent->>SkillsRuntime: Request package registry query guidance
alt Direct regxa usage
SkillsRuntime->>RegxaSkill: Load core regxa skill
Agent->>SkillsRuntime: Load PURL and API reference on demand
Agent->>Regxa: Run regxa operation with PURL
else AI SDK integration
SkillsRuntime->>AIToolSkill: Load AI tool integration skill
Agent->>SkillsRuntime: Load PURL cheatsheet on demand
Agent->>Regxa: Run package tool operation with PURL
end
Regxa->>Registry: Query package data across ecosystem
Registry-->>Regxa: Return package registry response
Regxa-->>Agent: Return normalized structured result
Generated by CodeAnt AI |
Adds agents/openai.yaml to both skills with interface metadata and policy config for Codex skill discovery.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 143ec97342
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
…dling description - Remove nonexistent per-ecosystem cache clear example - Clarify shorthand PURLs are CLI-only, library requires pkg: prefix - Document that packageTool throws exceptions, not structured errors
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
There was a problem hiding this comment.
2 issues found across 3 files (changes from recent commits).
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="skills/regxa-ai-tool/SKILL.md">
<violation number="1" location="skills/regxa-ai-tool/SKILL.md:115">
P2: This says the agent can self-correct after a tool error, but the examples here are single-step AI SDK calls. Without `stopWhen` or `ToolLoopAgent`, the model never gets another turn to recover from that failure.</violation>
</file>
<file name="skills/regxa/SKILL.md">
<violation number="1" location="skills/regxa/SKILL.md:30">
P2: This line overstates the contract. Only the PURL-based helpers require `pkg:` - the lower-level registry API later in this skill takes bare package names.</violation>
</file>
Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.
There was a problem hiding this comment.
10 issues found across 5 files
Confidence score: 3/5
- Some risk remains because
skills/regxa-ai-tool/SKILL.mdcurrently describespackageToolbehavior that does not matchsrc/ai.ts; without try/catch, invalid PURLs, not-found, or rate-limit errors can bubble up to AI SDK callers and cause user-facing failures. - Several high-confidence docs examples are inaccurate in
skills/regxa/SKILL.mdandskills/regxa/references/purl-guide.md(for example shorthand PURLs withoutpkg:for library APIs, unsupported cache command/options), which can lead to immediate runtime errors or incorrect CLI expectations. - This looks more like correctness/documentation drift than a broad code-architecture risk, so it may still be mergeable after targeted fixes to examples and claims.
- Pay close attention to
skills/regxa-ai-tool/SKILL.md,skills/regxa/SKILL.md, andskills/regxa/references/purl-guide.md- they contain the highest-impact mismatches that could mislead users into failing calls or non-working commands.
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="skills/regxa-ai-tool/SKILL.md">
<violation number="1" location="skills/regxa-ai-tool/SKILL.md:13">
P2: This install command is missing `zod`, which `regxa/ai` imports directly.</violation>
<violation number="2" location="skills/regxa-ai-tool/SKILL.md:26">
P2: This example calls `openai()` without importing it, so the streaming snippet does not compile as written.</violation>
<violation number="3" location="skills/regxa-ai-tool/SKILL.md:60">
P2: The documented dependency `scope` values are wrong here - regxa returns `development`, not `dev`, and it can also return `optional`.</violation>
<violation number="4" location="skills/regxa-ai-tool/SKILL.md:105">
P1: This is wrong - `packageTool` in `src/ai.ts` returns helper calls directly without a try/catch, so invalid PURLs, not-found, and rate-limit failures propagate as thrown errors to the AI SDK caller. Agents following this guidance will skip exception handling and crash on normal lookup failures. Should state that errors are thrown and callers need to handle them (or the AI SDK's error handling kicks in).</violation>
</file>
<file name="skills/regxa/references/purl-guide.md">
<violation number="1" location="skills/regxa/references/purl-guide.md:58">
P2: This makes shorthand sound universal, but only the CLI auto-prefixes `pkg:`. Library calls like `parsePURL("npm/lodash")` still throw.</violation>
</file>
<file name="skills/regxa/SKILL.md">
<violation number="1" location="skills/regxa/SKILL.md:27">
P2: This makes shorthand look valid for the library API, but only the CLI adds the missing `pkg:` prefix. Using `npm/lodash` with `fetch*FromPURL` will throw `InvalidPURLError`.</violation>
<violation number="2" location="skills/regxa/SKILL.md:65">
P2: `regxa cache clear npm` is a made-up command. `clear` only wipes the entire cache, so this example points agents at a feature that isn't there.</violation>
<violation number="3" location="skills/regxa/SKILL.md:68">
P2: This overstates the CLI. The query commands support `--json` and `--no-cache`, but the cache subcommands don't, so agents will expect JSON output that isn't implemented.</violation>
<violation number="4" location="skills/regxa/SKILL.md:104">
P2: `bulkFetchPackages` itself is not capped at 50. That limit only exists in the AI tool schema, so this comment teaches the wrong contract for the library helper.</violation>
<violation number="5" location="skills/regxa/SKILL.md:153">
P2: `registry` is undefined here. The example above created `npm`, so this snippet won't run if someone copies it as written.</violation>
</file>
Architecture diagram
sequenceDiagram
participant Agent as AI Agent (Vercel AI SDK)
participant Tool as NEW: regxa/ai (packageTool)
participant Lib as regxa (Library Core)
participant Cache as CachedRegistry (Disk/Mem)
participant Registry as External Registry (npm/PyPI/etc)
Note over Agent,Registry: Runtime Flow for regxa AI Tool Integration
Agent->>Tool: NEW: toolCall({ operation, purl })
Note right of Tool: Operations: package, versions,<br/>dependencies, maintainers, bulk
Tool->>Lib: NEW: parsePURL(purl)
Lib->>Lib: Normalize name (e.g. PyPI case/hyphen)
Lib-->>Tool: Parsed PURL Object
Tool->>Lib: NEW: fetch[Operation]FromPURL(purl)
Lib->>Cache: CHANGED: check cache entry
alt Cache Miss or --no-cache
Lib->>Registry: HTTP GET (Registry Endpoint)
Note right of Registry: Includes User-Agent & <br/>Exponential Backoff
alt 429 Rate Limited
Registry-->>Lib: Error
Lib->>Lib: Retry logic
else 200 OK
Registry-->>Lib: Raw Ecosystem JSON
end
Lib->>Lib: NEW: Map to normalized schema
Lib->>Cache: Store normalized result
else Cache Hit
Cache-->>Lib: Return cached data
end
Lib-->>Tool: Normalized Data Object
opt operation == "dependencies"
Tool->>Tool: Validate version presence in PURL
end
Tool-->>Agent: Structured Tool Result
Note left of Agent: Agent processes metadata<br/>or corrects PURL on error
Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (3)
skills/regxa/SKILL.md (1)
17-19: Missing language identifier on code block.Static analysis flagged this. Adding
textorplaintextsatisfies linters and helps syntax highlighters:Suggested fix
-``` +```text pkg:<type>/<name>@<version> -``` +```🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@skills/regxa/SKILL.md` around lines 17 - 19, Add a language identifier to the fenced code block that contains the package pattern (the block with "pkg:<type>/<name>@<version>") so linters and highlighters recognize it; update the opening fence from ``` to ```text (or ```plaintext) while leaving the content unchanged.skills/regxa/references/purl-guide.md (1)
7-9: Same language identifier issue.Consistency with the main SKILL.md would be nice:
Suggested fix
-``` +```text pkg:<type>/<namespace>/<name>@<version>?<qualifiers>#<subpath> -``` +```🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@skills/regxa/references/purl-guide.md` around lines 7 - 9, The code block containing the purl pattern "pkg:<type>/<namespace>/<name>@<version>?<qualifiers>#<subpath>" uses a plain triple-backtick fence; update the opening fence to include the language identifier by replacing ``` with ```text (leave the closing ``` as-is) so the block matches the same language tag used in SKILL.md and keeps rendering consistent.skills/regxa-ai-tool/SKILL.md (1)
91-95: MissingawaitonstreamText?Looking at Vercel AI SDK patterns,
streamTextreturns a promise. This might work because you're only accessingresult.textStreamlater, but the explicitawaitmatches line 28's pattern withgenerateTextand avoids confusion:const result = await streamText({Not a runtime bug since
textStreamaccess likely awaits internally, but inconsistent with thegenerateTextexample above.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@skills/regxa-ai-tool/SKILL.md` around lines 91 - 95, The assignment to result from streamText should be awaited for consistency with generateText and to ensure the Promise resolves before accessing result.textStream; update the call to await streamText(...) where result is declared (and make the enclosing function async if it isn’t), keeping the same options (model: openai("gpt-4o"), tools: { package: packageTool }, prompt: "...") so that result and result.textStream are accessed from the resolved value.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@skills/regxa-ai-tool/SKILL.md`:
- Around line 91-95: The assignment to result from streamText should be awaited
for consistency with generateText and to ensure the Promise resolves before
accessing result.textStream; update the call to await streamText(...) where
result is declared (and make the enclosing function async if it isn’t), keeping
the same options (model: openai("gpt-4o"), tools: { package: packageTool },
prompt: "...") so that result and result.textStream are accessed from the
resolved value.
In `@skills/regxa/references/purl-guide.md`:
- Around line 7-9: The code block containing the purl pattern
"pkg:<type>/<namespace>/<name>@<version>?<qualifiers>#<subpath>" uses a plain
triple-backtick fence; update the opening fence to include the language
identifier by replacing ``` with ```text (leave the closing ``` as-is) so the
block matches the same language tag used in SKILL.md and keeps rendering
consistent.
In `@skills/regxa/SKILL.md`:
- Around line 17-19: Add a language identifier to the fenced code block that
contains the package pattern (the block with "pkg:<type>/<name>@<version>") so
linters and highlighters recognize it; update the opening fence from ``` to
```text (or ```plaintext) while leaving the content unchanged.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: deb4f933-a731-48db-99dc-1860ca1677fc
📒 Files selected for processing (5)
skills/regxa-ai-tool/SKILL.mdskills/regxa-ai-tool/agents/openai.yamlskills/regxa/SKILL.mdskills/regxa/agents/openai.yamlskills/regxa/references/purl-guide.md
✅ Files skipped from review due to trivial changes (1)
- skills/regxa/agents/openai.yaml
📜 Review details
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: cubic · AI code reviewer
🧰 Additional context used
🧠 Learnings (16)
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : PURL input resolution and optional `pkg:` prefix normalization should be implemented in `src/commands/shared.ts`
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:38.679Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/core/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:38.679Z
Learning: No duplicate PURL parsing utilities outside purl.ts
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:54.862Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: test/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:54.862Z
Learning: Applies to test/unit/purl.test.ts : PURL contract tests should be located in `test/unit/purl.test.ts` for parse/build validation
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:38.679Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/core/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:38.679Z
Learning: PURL parse and build behavior is the single source of truth for validation and normalization
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Do not reimplement PURL parsing in individual command files; use the shared implementation from `src/commands/shared.ts`
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:03.586Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:03.586Z
Learning: Applies to src/**/*.ts : Do not parse PURLs outside `src/core/purl.ts`; call `createFromPURL` or `parsePURL` instead
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:12.605Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:12.605Z
Learning: Applies to src/{commands,helpers}/**/*.ts : Route all parsing logic through `createFromPURL` instead of duplicating in commands or helpers
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:12.605Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:12.605Z
Learning: Applies to src/helpers.ts : Wrap `createFromPURL` when adding convenience API in `src/helpers.ts` and preserve normalization path
Applied to files:
skills/regxa/references/purl-guide.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Do not mutate cache internals directly from commands; use the cache API instead
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:22.301Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/cache/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:22.301Z
Learning: Applies to src/cache/storage.ts : Storage lifecycle, global storage configuration, and disposal/clear hooks should be implemented in `src/cache/storage.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Shared argument flags (`--json` and `--no-cache`) should be defined and handled in `src/commands/shared.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:54.862Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: test/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:54.862Z
Learning: Applies to test/unit/{lockfile,cached-registry}.test.ts : Cache behavior tests should be located in `test/unit/lockfile.test.ts` and `test/unit/cached-registry.test.ts` for freshness, TTL, integrity, and wrapper behavior validation
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:22.301Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/cache/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:22.301Z
Learning: Applies to src/cache/cached-registry.ts : Cache wrapper behavior and read-through/write-through wrapper for `Registry` should be implemented in `src/cache/cached-registry.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:22.301Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/cache/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:22.301Z
Learning: Applies to src/cache/**/*.ts : Keep cache-key format stable; lockfile compatibility depends on maintaining a consistent key format
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:46.164Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/registries/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:46.164Z
Learning: Applies to src/registries/*.ts : Each registry adapter in `src/registries/` must expose `ecosystem`, `fetchPackage`, `fetchVersions`, `fetchDependencies`, `fetchMaintainers`, and `urls` exports
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:38.679Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/core/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:38.679Z
Learning: Applies to src/core/**/*.ts : Throw typed errors (InvalidPURLError, NotFoundError, RateLimitError) instead of plain Error in core flows
Applied to files:
skills/regxa-ai-tool/SKILL.md
🪛 markdownlint-cli2 (0.21.0)
skills/regxa/SKILL.md
[warning] 7-7: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
[warning] 19-19: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
[warning] 29-29: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
[warning] 38-38: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
[warning] 60-60: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
[warning] 80-80: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🔇 Additional comments (5)
skills/regxa-ai-tool/agents/openai.yaml (1)
1-12: Looks good.Clean manifest structure. The dependency on
regxatool andallow_implicit_invocation: falsepolicy are sensible defaults for a tool that should be explicitly invoked.skills/regxa/SKILL.md (1)
1-167: Solid documentation structure.Progressive disclosure pattern works well: quick-start CLI/library examples up front, references linked for deep dives. Error table at the end is helpful.
skills/regxa/references/purl-guide.md (1)
1-101: Comprehensive PURL reference.Good coverage of edge cases: scoped packages, encoding rules, normalization differences across ecosystems. The buildPURL/parsePURL examples showing encode/decode symmetry are helpful.
skills/regxa-ai-tool/SKILL.md (2)
1-115: Good AI tool integration guide.Clear progression: install → register → operations → streaming → errors. The discriminated union operation docs with JSON examples are exactly what an LLM agent would need to understand the tool schema.
21-26: Subpath export is correctly configured.The
regxa/aiimport path is properly set up in package.json with bothtypesandimportfields, so this won't break at runtime.> Likely an incorrect or invalid review comment.
- Fix dependency scope values (development not dev, add optional) - Add missing openai import in streaming example - Clarify maxSteps requirement for self-correction - Distinguish PURL helpers vs registry API for pkg: prefix - Remove incorrect 50-cap comment on bulkFetchPackages - Fix undefined registry variable in helper example - Limit --json/--no-cache docs to query commands only
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@skills/regxa/SKILL.md`:
- Around line 17-19: The fenced code block containing the literal text
"pkg:<type>/<name>@<version>" should include a language tag to satisfy
markdownlint (MD040); change the block start from ``` to ```text (or ```bash if
preferred) so it becomes ```text followed by pkg:<type>/<name>@<version> and
then ``` to close.
🪄 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: CHILL
Plan: Pro
Run ID: 8510f645-1108-426f-8c72-09ffd04a3f68
📒 Files selected for processing (2)
skills/regxa-ai-tool/SKILL.mdskills/regxa/SKILL.md
🚧 Files skipped from review as they are similar to previous changes (1)
- skills/regxa-ai-tool/SKILL.md
📜 Review details
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: cubic · AI code reviewer
🧰 Additional context used
🧠 Learnings (19)
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Do not mutate cache internals directly from commands; use the cache API instead
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:22.301Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/cache/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:22.301Z
Learning: Applies to src/cache/storage.ts : Storage lifecycle, global storage configuration, and disposal/clear hooks should be implemented in `src/cache/storage.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Shared argument flags (`--json` and `--no-cache`) should be defined and handled in `src/commands/shared.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:54.862Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: test/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:54.862Z
Learning: Applies to test/unit/{lockfile,cached-registry}.test.ts : Cache behavior tests should be located in `test/unit/lockfile.test.ts` and `test/unit/cached-registry.test.ts` for freshness, TTL, integrity, and wrapper behavior validation
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:22.301Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/cache/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:22.301Z
Learning: Applies to src/cache/cached-registry.ts : Cache wrapper behavior and read-through/write-through wrapper for `Registry` should be implemented in `src/cache/cached-registry.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:22.301Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/cache/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:22.301Z
Learning: Applies to src/cache/**/*.ts : Keep cache-key format stable; lockfile compatibility depends on maintaining a consistent key format
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : PURL input resolution and optional `pkg:` prefix normalization should be implemented in `src/commands/shared.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:03.586Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:03.586Z
Learning: Bulk fetch helpers should intentionally skip failed packages instead of failing all
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:46.164Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/registries/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:46.164Z
Learning: Applies to src/registries/*.ts : Each registry adapter in `src/registries/` must expose `ecosystem`, `fetchPackage`, `fetchVersions`, `fetchDependencies`, `fetchMaintainers`, and `urls` exports
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:46.164Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/registries/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:46.164Z
Learning: Use `npm.ts` as a reference implementation pattern when building new registry adapters
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:46.164Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/registries/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:46.164Z
Learning: Applies to src/registries/*.ts : Keep per-registry API quirks isolated to that adapter file before normalizing to shared types
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:03.586Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:03.586Z
Learning: Applies to src/registries/**/*.ts : Do not hardcode ecosystem registries using switch logic; use plugin-like registration factories instead
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Maintain parity between human-readable and `--json` output branches for new command output
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Keep command output in dual-mode supporting both human-readable and `--json` formats
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:29.354Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/commands/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:29.354Z
Learning: Applies to src/commands/**/*.ts : Do not reimplement PURL parsing in individual command files; use the shared implementation from `src/commands/shared.ts`
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:03.586Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:03.586Z
Learning: Applies to src/**/*.ts : Do not parse PURLs outside `src/core/purl.ts`; call `createFromPURL` or `parsePURL` instead
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:12.605Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:12.605Z
Learning: Applies to src/{commands,helpers}/**/*.ts : Route all parsing logic through `createFromPURL` instead of duplicating in commands or helpers
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:54.862Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: test/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:54.862Z
Learning: Applies to test/unit/purl.test.ts : PURL contract tests should be located in `test/unit/purl.test.ts` for parse/build validation
Applied to files:
skills/regxa/SKILL.md
📚 Learning: 2026-03-10T07:36:38.679Z
Learnt from: CR
Repo: oritwoen/regxa PR: 0
File: src/core/AGENTS.md:0-0
Timestamp: 2026-03-10T07:36:38.679Z
Learning: No duplicate PURL parsing utilities outside purl.ts
Applied to files:
skills/regxa/SKILL.md
🪛 markdownlint-cli2 (0.21.0)
skills/regxa/SKILL.md
[warning] 17-17: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
There was a problem hiding this comment.
2 issues found across 2 files (changes from recent commits).
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="skills/regxa-ai-tool/SKILL.md">
<violation number="1" location="skills/regxa-ai-tool/SKILL.md:116">
P2: This line is describing the wrong AI SDK behavior. Current AI SDK tool loops use `stopWhen`, and tool failures show up as `tool-error` parts instead of being thrown back to the caller in normal `generateText` or `streamText` usage.</violation>
</file>
<file name="skills/regxa/SKILL.md">
<violation number="1" location="skills/regxa/SKILL.md:155">
P3: This example uses `npm.urls()` without defining `npm`, so the snippet fails if someone copies it as shown.</violation>
</file>
Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Summary
Two agent skills following the agentskills.io spec for skills.sh:
packageToolinto Vercel AI SDK applications. Covers all 5 operations (package, versions, dependencies, maintainers, bulk).Both skills use progressive disclosure - lean SKILL.md files with JiT-loaded references for detailed type contracts, PURL encoding rules, and common mistakes.
Test plan
namefields