Skip to main content
Agent skill

Plugin Sdk Impact

Count Plugin SDK consumers across live ClawHub releases and bundled OpenClaw plugins before deprecation, removal, or behavior changes; supports public release-date filters.

Get started

Get the complete skill folder, including its scripts and reference files.

When to use this skill

Use Plugin Sdk Impact when an AI agent needs a reusable SKILL.md workflow for this job: Count Plugin SDK consumers across live ClawHub releases and bundled OpenClaw plugins before deprecation, removal, or behavior changes; supports public release-date filters.

When not to use it

Skip Plugin Sdk Impact when the task is outside the coding category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.

How to install

Where will you use it?

Install scope
Claude Code folder setup
mkdir -p ~/.claude/skills/plugin-sdk-impact
# Save SKILL.md to ~/.claude/skills/plugin-sdk-impact/SKILL.md
  1. Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/plugin-sdk-impact/SKILL.md. For a shared project, commit the folder at .claude/skills/plugin-sdk-impact/ instead.
  2. Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
  3. For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.

Full install guide for Claude, Cursor, and Codex →

Try a starter prompt

After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.

Starter prompt
Use the Plugin Sdk Impact skill for this task: Use Plugin Sdk Impact when this documented workflow matches the task. Confirm any required inputs or files before making changes.

What supports this recommendation?

71/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.

Documentation
77/100
Practical detail
68/100
Source evidence
63/100
Source signals
72/100
Model compatibility evidence
  • claudeuntested

    No model-specific signal or recorded compatibility test was found.

  • chatgptuntested

    No model-specific signal or recorded compatibility test was found.

  • geminiuntested

    No model-specific signal or recorded compatibility test was found.

  • copilotuntested

    No model-specific signal or recorded compatibility test was found.

  • llamauntested

    No model-specific signal or recorded compatibility test was found.

  • perplexityuntested

    No model-specific signal or recorded compatibility test was found.

  • mistraluntested

    No model-specific signal or recorded compatibility test was found.

  • grokuntested

    No model-specific signal or recorded compatibility test was found.

Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.

What this skill does

# Plugin SDK impact

Combine current ClawHub inventories with a source scan of every bundled OpenClaw plugin, including plugins absent from ClawHub, to inform compatibility decisions. Static usage identifies dependencies; it does not prove execution or reliance on a particular behavior. Keep the approval and deprecation rules in [the SDK boundary guide](../../../src/plugin-sdk/AGENTS.md) authoritative.

## Investigate

1. Identify every affected public surface from source, including alternate SDK exports and runtime paths. Verify spelling against the definition and known callers; an incorrect key can return a convincing zero. Trace grouped facades, SDK helpers, and registration readers as well: a flat alias can have zero direct plugin calls while implementing a supported grouped method. 2. Scan every bundled plugin in a clean OpenClaw checkout using Plugin Inspector's static source analysis. Reuse an installed Inspector 0.4.3 or newer; otherwise install it in a separate temporary directory, without lifecycle scripts:

```sh inspector_dir=$(mktemp -d) npm install --prefix "$inspector_dir" --ignore-scripts @openclaw/[email protected] node {baseDir}/scripts/scan-bundled.mjs \ --inspector "$inspector_dir/node_modules/@openclaw/plugin-inspector" \ --out /tmp/openclaw-bundled-sdk-usage.json ```

The scanner defaults to this skill's checkout; `--root <checkout>` selects another. It discovers `extensions/*/openclaw.plugin.json`, never executes plugin code, and records the commit, input fingerprint, Inspector version, every plugin, and incomplete analyses. A failed or timed-out scan remains a gap. Plugins without declared entrypoints also remain coverage gaps, because Inspector may miss implicit entrypoints. An interrupted corpus scan cannot be used as a finished inventory. Reuse the report for multiple queries while its source remains unchanged. A changed commit, plugin roster, or uncommitted source invalidates it; rescan instead of editing the receipt.

3. Query each canonical key with both sources:

```sh node {baseDir}/scripts/query.mjs 'api-member:api.registerTool' \ --bundled /tmp/openclaw-bundled-sdk-usage.json ```

For “updated within the last three months”, add `--months 3`. This means the latest eligible public release was uploaded within three calendar months of now, measured in UTC (month ends are clamped). Listing edits and rescans do not count as updates. The stored release date is its creation/upload time; public approval can happen later. `--since 2026-07-16T12:00:00Z` accepts an explicit cutoff. Report the exact inclusive cutoff from `releaseWindow`, not a guessed date. Public consumers and coverage use the same window on every page. Bundled source has no comparable upload date: report it separately, unfiltered, even for a recent-release question. The combined count is the union of that public release window and current bundled source, not a count of recent releases.

The helper follows all usage and coverage pages, including empty intermediate pages. Counts and coverage include every result; `items`, `gaps`, and `warnings` contain bounded examples with omitted counts in `output.omitted`. Mixed references count in both runtime and type totals. For an investigation needing every consumer or gap, save `--full` output to a file and read it in sections:

```sh node {baseDir}/scripts/query.mjs 'api-member:api.registerTool' \ --bundled /tmp/openclaw-bundled-sdk-usage.json --full > sdk-usage.json ```

`--help` documents local endpoint and traversal limits.

4. Inspect matched evidence against the proposed change. A method call may not use the behavior being changed. Type-only consumers still matter for type/export removal. Query related members separately; a module import does not prove use of every export. Repeated references in a plugin count as one consumer. 5. For a usage-count question, report the key, query time, exact release window (if requested), ClawHub and bundled counts, their combined count, bundled commit, and both coverage gaps. Top-level `distinctPlugins` and `coverage` remain ClawHub-only; `bundled` describes source results; `combined` merges exact package identities once. Some bundled plugins have no package name: when they could overlap public matches, `combined.distinctPlugins` is null and `distinctPluginBounds` reports the possible range. `observedPluginIdentities` counts source identities before that uncertain overlap. Report the range, not its upper bound as an exact count. Different names remain distinct; never guess aliases from display names. Missing bundled evidence makes combined impact unavailable. For a proposed SDK change, also report matched evidence, indirect callers, and your compatibility recommendation. Finish change investigations with the maintainer's decision: preserve behavior, deprecate with migration, or proceed under existing policy. Never infer permission from the count.

## Canonical keys

| Surface | Example | | ------------------- | -------------------------------------------------------------- | | SDK module | `sdk-module:openclaw/plugin-sdk/core` | | Named SDK export | `sdk-export:openclaw/plugin-sdk/core#OpenClawPluginApi` | | Plugin API method | `api-member:api.registerTool` | | Host runtime member | `runtime-member:runtime.channel.reply.dispatchReplyFromConfig` | | Named hook | `hook:hook.before_agent_start` | | Manifest contract | `manifest-contract:manifest.contracts.tools:example_tool` |

Keys describe SDK identities, not local aliases. `api.runtime` members normalize to `runtime.*`. The inventory includes supported and deprecated references; nonempty `deprecationCodes` selects entries associated with existing warnings. Per-plugin inspection is available at `https://clawhub.ai/api/v1/sdk-usage/plugin?package=<encoded-name>`; follow its `nextCursor` until null to obtain the full inventory.

## Interpret the result

| Result | Report | | --------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Observed consumers | Report public, bundled, and combined counts with their separate scopes. Inspect evidence before judging a behavior change. | | No observed consumers, complete coverage | “No static references detected across N public/bundled plugin identities.” Requires both scans; excludes private non-bundled plugins and unbundled dependencies. Not proof that nobody uses the API. | | Partial, failed, missing, incompatible, or unresolved analysis | State observed matches, gap counts, and representative gaps. Zero matches leaves impact unknown. | | Empty corpus | No eligible plugins were observed; this provides no ecosystem assurance. | | Unavailable endpoint, malformed result, or incomplete traversal | Impact is unknown. Preserve the error and use local evidence; never substitute zero. |

Coverage is tied to current release/artifact and analysis versions. Eligible ClawHub reports have `inputKind: "artifact"`; bundled reports use `"source"`. A source-checkout benchmark cannot establish what a published package contains. Source-only type references still matter for SDK type and export removal; preserve the source of each match. An old scan timestamp alone does not make an immutable artifact stale. Evidence is sampled; each scan reports its evidence/unknown limits. Pages are live reads, not one snapshot: rerun if releases or visibility change during the investigation. The helper flags changes it observes but cannot establish a transaction-wide snapshot.

Do not dismiss a security issue just because no consumers were observed. Describe how usage evidence affects compatibility risk; security and SDK owners still decide the fix and any exception to the normal migration window.

Intended uses

  • Use Plugin Sdk Impact when this documented workflow matches the task.

Implementation notes

  • Review the source instructions and adapt inputs before running the workflow.

Try it, then tell us how it went

A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.