feat(genkit): add genkit plugin bundling official genkit-ai/skills - #280
Merged
Merged
Conversation
Bundle the 4 official Genkit skills from github.com/genkit-ai/skills (developing-genkit-js, -go, -dart, -python) via skills.sh with a pinned skills-lock.json, so the weekly update-skills workflow tracks upstream. Registered in the Claude, Codex, and Cursor marketplaces (with a relevance block keyed on the genkit CLI and Genkit manifest deps) and in release-please.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Up to standards ✅🟢 Issues
|
🔍 Tessl Skill Review
|
| Dimension | Score | Detail |
|---|---|---|
| conciseness | ████░ 4/5 | The body is dense with non-obvious, Dart-specific facts (experimental import entry points, interrupts-as-tools, the lazy top-level final gotcha, --non-interactive and --format json flags) and avoids explaining concepts Claude already knows. It falls short of anchor 5 because passages like the Agents prose paragraph ("The agent/session/snapshot APIs are experimental and live behind opt-in imports...") and the multi-sentence CLI behavior notes pack background explanation that could be trimmed or pushed to a reference. |
| actionability | █████ 5/5 | Commands are copy-paste ready and cover the common cases: install (curl -sL cli.genkit.dev | bash), run (genkit start -- dart run main.dart, --noui, --non-interactive variants), one-shot flow runs (genkit flow:run myFlow '{"data": "input"}' -- dart run main.dart), trace inspection (genkit trace:get <traceId> --format json), docs lookup, and validation (dart analyze). This matches anchor 5 ('fully executable; copy-paste ready code or commands; specific examples cover the common cases'). |
| workflow clarity | ████░ 4/5 | Validation and feedback loops are present — "Always check that code cleanly compiles using dart analyze before generating the final response", trace-based verification ("Verify with traces, not a blind run"), and error recovery for the concrete failure "Process exited before runtime was ready" with its fix (reference symbols from main()). It stops short of anchor 5 because the overall workflow (choose agent vs flow → implement via references → run under genkit start → verify traces → dart analyze) is scattered across sections rather than presented as one explicit sequence with checkpoints. |
| progressive disclosure | ███░░ 3/5 | The structure itself is strong — one-level-deep references, clearly signaled with per-topic descriptions and a plugin table of "Load for..." rows — but references/agents.md is referenced three times (the "start here" Agents bullet, the flow:run note, and Best Practices) and does not exist in the bundle, so navigation fails at the primary entry point for the skill's largest feature area. Working references with a broken critical one is 'some structure but could be better organized' (anchor 3) rather than anchor 4's 'minor organization gaps'; without the dead link this would score 5. |
Suggestions:
- Fix the broken reference:
references/agents.mdis cited as the 'start here' entry for the Agents section (and twice more) but does not exist in the bundle — either add the file or repoint those links to an existing reference such as agents-custom.md. - Tighten the Agents section: the ~15-line prose paragraph mixing experimental-import warnings, client interop, and middleware notes could become a short bullet list with the import/warning details moved into the agents reference.
- Make the end-to-end workflow explicit: a short ordered sequence (pick agent vs flow → implement per reference → run under
genkit start→ verify withgenkit trace:get→dart analyze) at the top of Best Practices would turn scattered guidance into a single checklist.
Description Review
| Dimension | Score | Detail |
|---|---|---|
| specificity | ███░░ 3/5 | "Generates code and provides documentation for the Genkit Dart SDK" names the domain and two actions, but those actions are generic — it never says what kinds of code (agents, flows, tools, prompts) or what the documentation covers. This matches anchor 3 ('names domain and 1-2 concrete actions, but not comprehensive'); it is above anchor 2 because the domain is precisely named, and below anchor 4 because the actions themselves lack concreteness compared to 'Extracts text from PDF files, fills forms, converts pages to images'. |
| completeness | █████ 5/5 | Both halves are explicit: the 'what' is stated directly ("Generates code and provides documentation for the Genkit Dart SDK") and the 'when' is an explicit "Use when the user asks to..." clause with three concrete trigger scenarios. This matches anchor 5 ('clearly and explicitly answers both what AND when with concrete trigger phrases'); anchor 4 does not fit because its caveat ('when' could be more explicit or specific) does not apply — the when-clause is already specific. |
| trigger term quality | ████░ 4/5 | The trigger clause surfaces natural user phrasing: "build AI agents in Dart", "use Genkit flows", "integrate LLMs into Dart/Flutter applications", plus "Genkit Dart SDK" itself — a user needing this skill would plausibly say most of these. It falls short of anchor 5's comprehensive synonym coverage: common phrasings like prompts/Dotprompt, chatbot, or model plugin names (Gemini, Anthropic, OpenAI in a Dart context) are absent, so a few natural terms are missing, which is exactly anchor 4. |
| distinctiveness conflict risk | █████ 5/5 | "Genkit Dart SDK" combined with Dart/Flutter-specific triggers carves out a clear niche with minimal overlap; a user mentioning Genkit, Dart, or Flutter LLM integration is almost certainly asking for this skill. It matches anchor 5 ('clear niche with distinct triggers; minimal conflict risk'); the only conceivable overlap would be a hypothetical Genkit TypeScript/Go skill, which is not a present-tense conflict. |
plugins/genkit/.agents/skills/developing-genkit-go/SKILL.md
A well-structured, highly actionable reference skill: complete executable Hello World, precise CLI/debugging commands, and a clean routing table that keeps the body lean while deferring depth to 17 real bundle files. The two real weaknesses are a broken 'start here' reference (references/agents.md is missing) and modest padding in the CLI section.
Validation
⚠️ metadata_version — 'metadata.version' is missing⚠️ relative_links — Relative link issues: 1 missing⚠️ referenced_paths_exist — Referenced path issues: 1 missing
Full review details
Validation Checks
13/16 checks passed.
Review Details
| Dimension | Score | Detail |
|---|---|---|
| conciseness | ████░ 4/5 | Nearly every token is Genkit-specific knowledge Claude would not have (experimental gating, genkit start semantics, --format json guidance, middleware concurrency caveats), and the Hello World block is complete without commentary. Minor padding keeps it from 5 — e.g. "unintrusively wraps any Go program... running it unchanged" and "That is expected and correct for the common cases" could each be tightened by roughly half. |
| actionability | █████ 5/5 | The Hello World example is fully executable copy-paste Go (imports, init, flow definition, HTTP serving), the CLI section gives exact runnable commands with flags (genkit start -- go run ., flow:run ... --stream, trace:get <id> --format json, install one-liner), and Key Guidance gives concrete patterns like jsonschema:"description=..." struct tags. Not a 4: no significant gaps remain — common cases (build, run, debug, look up docs) are all covered with ready-to-run commands. |
| workflow clarity | ████░ 4/5 | A clear path exists — Hello World → route to the right reference per need → run under genkit start (not blind go run .) → inspect traces → docs:search for specifics — with an explicit verification loop ("Verify with traces, not a blind run" backed by trace commands). Not a 5 because the sequence is implied by section order rather than presented as one explicit workflow with checkpoints, and the agents path stops at a routing list. Not a 3: validation guidance is present and concrete, not merely implicit. |
| progressive disclosure | ████░ 4/5 | The body is a genuine overview: a feature→reference routing table with 'When to load' guidance, a one-level-deep reference set, and detailed material correctly deferred to bundle files. However, the Agents section's 'start here' entry point references/agents.md does not exist in the bundle (verified against the file listing) — a broken primary link in an otherwise excellent navigation structure, which is a minor-to-material organization gap consistent with the 4 anchor rather than the 5 anchor's 'easy navigation'. |
Suggestions:
- Create or restore references/agents.md — the Agents section marks it as 'start here' but the file is absent from the bundle, breaking the primary navigation path into the nine agent sub-references.
- Tighten the Genkit CLI prose: e.g. compress 'unintrusively wraps any Go program... running it unchanged while capturing traces' and the 'That is expected and correct for the common cases' paragraph to single clauses; the commands already carry the meaning.
- Make the core workflow explicit as a short sequenced list (init → define flow → serve → run under
genkit start→ inspect traces) so the verification loop is a stated checklist rather than implied by section order.
Description Review
| Dimension | Score | Detail |
|---|---|---|
| specificity | ████░ 4/5 | "Develop AI-powered applications using Genkit in Go" plus an explicit capability list ("generation, prompts, streaming, tool calling, or model providers") names the domain and several concrete actions. Falls short of a 5 because the actions are feature-area names rather than fully concrete operations like the anchor's "fill forms, merge documents", and a couple of core areas (flows, middleware) only appear in the trigger clause. |
| completeness | █████ 5/5 | Explicitly answers both: what ("Develop AI-powered applications using Genkit in Go") and when ("Use when the user asks to build AI features, agents, flows, or tools in Go using Genkit, or when working with Genkit Go code involving...") with concrete trigger phrases, matching the top anchor. Score 4 would require the 'when' to be less explicit, which it is not. |
| trigger term quality | ████░ 4/5 | Natural phrases a user would say are well covered: "build AI features", "agents", "flows", "tools", "in Go", "Genkit", "streaming", "tool calling". Not a 5 because common synonyms users might say — "chatbot", "LLM app", "Gemini", "Vertex" — are absent from the trigger list. |
| distinctiveness conflict risk | █████ 5/5 | "in Go using Genkit" carves out a clear niche with distinct triggers — Go language plus the Genkit SDK — leaving minimal conflict risk; even against a sibling Genkit JS skill the Go qualifier disambiguates. It is not below-5 material since the triggers are unambiguous and specific. |
plugins/genkit/.agents/skills/developing-genkit-js/SKILL.md
The body is highly actionable and workflow-oriented, with executable examples, explicit validation steps, and a clear split between inline overview and reference files. Its two real defects are the broken references/agents.md link (the designated starting point for agents) and redundant repetition of the error-consultation protocol across multiple sections.
Validation
⚠️ metadata_version — 'metadata.version' is missing⚠️ relative_links — Relative link issues: 4 missing⚠️ referenced_paths_exist — Referenced path issues: 4 missing
Full review details
Validation Checks
13/16 checks passed.
Review Details
| Dimension | Score | Detail |
|---|---|---|
| conciseness | ███░░ 3/5 | The body is largely dense and useful, but the 'read common-errors.md first' instruction is repeated in at least four places ("Critical: Do Not Trust Internal Knowledge", the numbered Error Troubleshooting Protocol with its DO NOT list and "non-negotiable" emphasis, and workflow steps 5-6), which is unnecessary duplication. Matches anchor 3 (mostly efficient but could be tightened); not 4 because the repeated protocol sections are more than minor trimming, not 2 because there is no padding that explains concepts Claude already knows. |
| actionability | █████ 5/5 | Fully executable throughout: a complete Hello World flow, exact install/upgrade commands with versions ("npm install -g genkit-cli@^1.29.0"), and copy-paste CLI patterns ("genkit start -- npx tsx --watch src/index.ts", "genkit flow:run myFlow '{"data": "input"}' -- npx tsx src/index.ts", "genkit trace:get --format json") with concrete gotchas (flow:run does not fall back to schema defaults, --noui is not one-shot). Matches anchor 5; not 4 because the examples cover the common cases with no missing key details. |
| workflow clarity | █████ 5/5 | The Development Workflow is a clearly sequenced decision process (agent vs flow, provider default, framework detection, best practices, correctness, error handling) with explicit validation checkpoints ("Run type checks (e.g., npx tsc --noEmit)", "Verify with traces, not a blind run") and a feedback loop (match error to common-errors.md, apply fix, fall back to genkit docs:search only if not found). Matches anchor 5 with explicit validation, recovery loops, and the troubleshooting checklist; not 4 because no material checkpoint is missing. |
| progressive disclosure | ███░░ 3/5 | Structure is otherwise good (topic sections each pointing to one-level-deep, clearly labeled reference files, plus a consolidated References list), but references/agents.md is cited four times — including as the "start here" entry point for agents and in the References index — and does not exist in the bundle, so the primary navigation path for the skill's flagship topic is broken. Not 4 because a missing core reference is more than a minor organization gap; not 2 because all 18 other references are real, well-signaled, and one level deep. |
Suggestions:
- Add the missing references/agents.md file (or repoint its four references, including the "start here" link and the References index entry, to an existing file such as agents-custom.md or a merged agents doc).
- State the 'read common-errors.md before any fix' rule once (e.g., in the Error Troubleshooting Protocol) and reference it from the Development Workflow instead of repeating the full protocol in three sections.
- Trim the emphasized DO NOT list and "non-negotiable" framing, which restates the same mandatory-first-step instruction already given above it.
Description Review
| Dimension | Score | Detail |
|---|---|---|
| specificity | ███░░ 3/5 | "Develop AI-powered applications using Genkit in Node.js/TypeScript" names the domain and a single development action, but does not list concrete capabilities like defining flows, building agents, or writing .prompt files. It matches anchor 3 (domain plus 1-2 concrete actions, not comprehensive); not 4 because no several specific actions are enumerated, not 2 because the domain and action are explicit rather than generic. |
| completeness | █████ 5/5 | Explicitly answers both: what ("Develop AI-powered applications using Genkit in Node.js/TypeScript") and when ("Use when the user asks about Genkit, AI agents, flows, or tools... or when encountering Genkit errors...") with multiple concrete trigger phrases. Matches anchor 5; not 4 because the 'when' clause is already explicit and specific rather than needing more detail. |
| trigger term quality | ████░ 4/5 | Triggers include natural terms users would say: "asks about Genkit, AI agents, flows, or tools in JavaScript/TypeScript", plus error-oriented phrases ("validation issues, type errors, or API problems"). Not 5 because a few natural synonyms are missing (e.g., prompts/dotprompt, plugins, deployment); not 3 because coverage goes beyond a couple of generic keywords with common variations included. |
| distinctiveness conflict risk | ████░ 4/5 | "Genkit" is a clear niche trigger with minimal conflict risk, but the second trigger clause ("validation issues, type errors, or API problems") is not explicitly Genkit-scoped and could fire on generic JS type errors, giving minor overlap with general troubleshooting skills. Not 5 due to that overlap risk; not 3 because the dominant Genkit keyword keeps it mostly distinct. |
Suggestions:
- Enumerate 2-3 concrete capabilities in the 'what' clause (e.g., 'define flows, build multi-turn agents, write .prompt files') to raise specificity.
- Scope the error triggers to Genkit (e.g., 'Genkit validation or type errors') to reduce overlap with generic JavaScript troubleshooting skills.
plugins/genkit/.agents/skills/developing-genkit-python/SKILL.md
A tight, highly actionable overview with an explicit verify-then-debug workflow and a well-organized one-level-deep reference bundle. Two issues hold it back: a prominent broken reference (agents.md, the 'Start here' doc for the featured capability) and some duplicated guidance between the Workflow and CLI sections.
Validation
⚠️ metadata_version — 'metadata.version' is missing⚠️ relative_links — Relative link issues: 4 missing⚠️ referenced_paths_exist — Referenced path issues: 4 missing
Full review details
Validation Checks
13/16 checks passed.
Review Details
| Dimension | Score | Detail |
|---|---|---|
| conciseness | ████░ 4/5 | The body is lean and assumes competence (complete Hello World, terse import list, command-first CLI section), with no explanations of concepts Claude already knows. Not 5: there is real duplication — 'Running the app directly (uv run) does not capture dev traces' appears in both Workflow step 5 and the CLI section, and the --format json paragraph repeats the comment already shown in the trace-commands code block. |
| actionability | █████ 5/5 | Fully executable, copy-paste-ready guidance throughout: a complete runnable Hello World program, exact CLI invocations (genkit start -- uv run src/main.py, --non-interactive, flow:run with JSON input), and concrete trace-inspection commands covering the common run/verify/debug cases. The minor gap is that the agent snippet relies on a reference file for completion. |
| workflow clarity | █████ 5/5 | The six-step Workflow is clearly sequenced with an explicit decision rule (agent vs. flow), environment setup, entry-point guidance, an explicit verification step ('Verify with traces, not a blind run') with the commands to do it, and an error-recovery step ('Stuck? Common Errors first'). This matches the anchor with explicit validation steps and a feedback loop for recovery. |
| progressive disclosure | ████░ 4/5 | Clear hub-and-spoke overview: 14 of 15 referenced files exist, all exactly one level deep with no nested references, clearly signaled inline and in a References section. Not 5: references/agents.md is missing from the bundle despite being linked four times — including as the 'Start here' entry point for the headline Agents feature — which breaks navigation for that entire topic; still above anchor 3 because structure and signaling are otherwise exemplary and nothing that belongs in a separate file is inlined. |
Suggestions:
- Add the missing references/agents.md (or repoint its four links to the existing agents-* pages), since it is the 'Start here' entry for the skill's headline feature and currently dead-ends.
- Deduplicate the trace-capture warning: state 'running directly with
uv runskips trace capture' once and cross-reference it from the other location. - Trim the paragraph after the trace commands that re-explains
--format json, which the code-block comment already conveys.
Description Review
| Dimension | Score | Detail |
|---|---|---|
| specificity | ███░░ 3/5 | "Develop AI-powered applications using Genkit in Python" names the domain and one action, but the action is generic and the concrete capabilities the skill covers (generate, stream, tools, flows, agents) are absent. Not 4: it does not list several specific actions; not 2: it goes beyond naming the domain with a stated purpose and platform. |
| completeness | █████ 5/5 | Explicitly answers what ("Develop AI-powered applications using Genkit in Python") and when ("Use when the user asks about Genkit, AI agents, flows, or tools in Python, or when encountering Genkit errors, import issues, or API problems") with concrete trigger phrases, matching the top anchor. |
| trigger term quality | ████░ 4/5 | "Genkit, AI agents, flows, or tools in Python" plus "Genkit errors, import issues, or API problems" are phrases users would naturally say. Not 5: a few natural entry points (e.g., Genkit CLI, Dev UI, dotprompt, streaming) that the body covers are missing from the trigger list. |
| distinctiveness conflict risk | ████░ 4/5 | Genkit is a clear niche framework with framework-pinned triggers, giving minimal conflict risk. Not 5: "AI agents ... in Python" could also fire for non-Genkit agent work in closely related Python agent-building skills. |
Suggestions:
- Name the concrete capabilities in the description (e.g., 'Generate, stream, and structure model output; build flows, tools, and multi-turn agents with Genkit in Python') to raise specificity from a single generic action.
- Consider adding 'Genkit CLI' or 'genkit start' to the trigger clause, since running and debugging with the CLI/Dev UI is a major focus of the skill body.
To improve your score, point your agent at the Tessl optimization guide. Need help? Jump on our Discord.
Feedback
Report issues with this review at tesslio/skill-review, or send private feedback from your terminal with tessl feedback.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request introduces the official Genkit plugin, adding support for Genkit skills across JavaScript/TypeScript, Go, Dart/Flutter, and Python. It updates the marketplace manifests for Claude, Cursor, and other platforms, configures release-please, and vendors the respective documentation/skills under plugins/genkit/.agents/skills/. A review comment suggests improving the regex pattern for pubspec.yaml in .claude-plugin/marketplace.json by replacing the leading newline character with a line-start anchor to ensure robust line-by-line matching.
amondnet
marked this pull request as ready for review
September 24, 2026 16:35
Contributor
Author
|
/gemini review |
Merged
Contributor
There was a problem hiding this comment.
Code Review
This pull request adds the official Genkit plugin to the marketplace, enabling support for building AI-powered applications with Genkit in JavaScript/TypeScript, Go, Dart/Flutter, and Python. The changes register the plugin in the marketplace configuration files for Claude, Cursor, and other agents, update the release manifests, add documentation, and vendor the corresponding development skills under the .agents/skills/ directory. Since there are no review comments, I have no additional feedback to provide.
|
 |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
| "flutter", | ||
| "python" | ||
| ], | ||
| "skills": "./.agents/skills/" |
There was a problem hiding this comment.
When someone installs Genkit through the Codex marketplace, the plugin has no loadable skills. Its only four skills are under ./.agents/skills/, but the Codex generator omits the skills field for that path, and there is no root skills/ directory. The plugin installs without making its skills available. Give Codex a compatible skill path or layout.
Prompt To Fix With AI
This is a comment left during a code review.
Path: plugins/genkit/.claude-plugin/plugin.json
Line: 26
Comment:
**Codex cannot load Genkit skills**
When someone installs Genkit through the Codex marketplace, the plugin has no loadable skills. Its only four skills are under `./.agents/skills/`, but the Codex generator omits the `skills` field for that path, and there is no root `skills/` directory. The plugin installs without making its skills available. Give Codex a compatible skill path or layout.
---
For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.
| }, | ||
| { | ||
| "file": "[/\\\\]pubspec\\.yaml$", | ||
| "pattern": "(^|\\n)\\s*genkit(_[a-z_]+)?\\s*:" |
There was a problem hiding this comment.
The dependency pattern allows only letters and underscores after genkit_, so it does not match the Genkit Dart package genkit_a2ui. A Dart project that declares this package without another matching Genkit dependency will miss the plugin suggestion. Allow digits in the package-name suffix.
Suggested change
| "pattern": "(^|\\n)\\s*genkit(_[a-z_]+)?\\s*:" | |
| "pattern": "(^|\\n)\\s*genkit(_[a-z0-9_]+)?\\s*:" |
Prompt To Fix With AI
This is a comment left during a code review.
Path: .claude-plugin/marketplace.json
Line: 1356
Comment:
**Dart signal misses genkit_a2ui**
The dependency pattern allows only letters and underscores after `genkit_`, so it does not match the Genkit Dart package `genkit_a2ui`. A Dart project that declares this package without another matching Genkit dependency will miss the plugin suggestion. Allow digits in the package-name suffix.
```suggestion
"pattern": "(^|\\n)\\s*genkit(_[a-z0-9_]+)?\\s*:"
```
---
For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
| "flutter", | ||
| "python" | ||
| ], | ||
| "skills": "./.agents/skills/" |
There was a problem hiding this comment.
The four skills live only under .agents/skills/. Claude uses this declared path, but the generated Codex manifest omits its skills field because Codex requires a root skills/ directory. Cursor also discovers skills from a default folder at the plugin root. As a result, installing the advertised Codex or Cursor plugin does not make the Genkit skills available. Provide a supported skills layout for those runtimes.
Prompt To Fix With AI
This is a comment left during a code review.
Path: plugins/genkit/.claude-plugin/plugin.json
Line: 26
Comment:
**Skills unavailable in Codex and Cursor**
The four skills live only under `.agents/skills/`. Claude uses this declared path, but the generated Codex manifest omits its `skills` field because Codex requires a root `skills/` directory. Cursor also discovers skills from a default folder at the plugin root. As a result, installing the advertised Codex or Cursor plugin does not make the Genkit skills available. Provide a supported skills layout for those runtimes.
---
For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
Comment on lines
+35
to
+36
| router.post('/api/$path/getSnapshot', shelfHandler(agent.getSnapshotDataAction)); | ||
| router.post('/api/$path/abort', shelfHandler(agent.abortAgentAction)); |
There was a problem hiding this comment.
If an application deploys this example, it mounts snapshot-reading and abort actions without authentication or a snapshot-ownership check, then serves them on all IPv4 interfaces with permissive CORS. A caller who obtains another user's snapshot ID can read that conversation's messages and artifacts or cancel its background work. Protect these routes with server-side authentication and ownership checks.
How this was verified: The example exposes both actions through unguarded POST routes, while its snapshot examples show that an ID retrieves conversation state and controls background tasks.
Prompt To Fix With AI
This is a comment left during a code review.
Path: plugins/genkit/.agents/skills/developing-genkit-dart/references/agents-deployment.md
Line: 35-36
Comment:
**Snapshot endpoints lack access checks**
If an application deploys this example, it mounts snapshot-reading and abort actions without authentication or a snapshot-ownership check, then serves them on all IPv4 interfaces with permissive CORS. A caller who obtains another user's snapshot ID can read that conversation's messages and artifacts or cancel its background work. Protect these routes with server-side authentication and ownership checks.
**How this was verified:** The example exposes both actions through unguarded POST routes, while its snapshot examples show that an ID retrieves conversation state and controls background tasks.
---
For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.This branch was successfully deployed
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
Adds a new
genkitplugin that bundles the four official skills from genkit-ai/skills:developing-genkit-jsdeveloping-genkit-godeveloping-genkit-dartdeveloping-genkit-pythonChanges
plugins/genkit: skills installed viabunx skills add ... --agent universaland pinned inplugins/genkit/skills-lock.json, so the weeklyupdate-skillsworkflow tracks upstream."skills": "./.agents/skills/". Codex, Cursor, and Antigravity manifests were generated withbun scripts/cli.ts multi-format. I reverted the unrelated repo-wide churn that command produced..claude-plugin/marketplace.jsonwith arelevanceblock:genkitpackage.json(genkit/@genkit-ai/*),go.mod(github.com/firebase/genkit/go),pubspec.yaml(genkit*),pyproject.toml/requirements*.txt(genkit*)release-please-config.jsonand.release-please-manifest.jsonentries (1.0.0) covering all 4 version-bearing manifests.Reviewer notes
NOASSERTIONbecausegenkit-ai/skillshas no LICENSE file. The Genkit framework itself is Apache-2.0.plugins/genkit/.agents/skills/are vendor-synced. Please do not review or edit them.Related issue
None.
Checklist
claude plugin validate plugins/genkitpassedclaude plugin validate .claude-plugin/marketplace.jsonpassed (the only warnings are about missingauthoron existing plugins)Summary by cubic
Adds a new
genkitplugin that bundles the four official skills from genkit-ai/skills covering Genkit development in JavaScript/TypeScript, Go, Dart/Flutter, and Python.Notes
plugins/genkit/skills-lock.jsonso the weeklyupdate-skillsworkflow tracks upstream; files under.agents/skills/are vendor-synced and should not be reviewed.genkitCLI and Genkit manifest dependencies; the Dart relevance matches the genkit dependency on the first line ofpubspec.yaml.NOASSERTIONbecausegenkit-ai/skillshas no LICENSE file.Written for commit 32835b0. Summary will update on new commits.