AGENT SKILL

plugin-lifecycle

Create, inspect, install, catalog, add, and build Solace Agent Mesh plugin packages safely.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
plugin-lifecycle
license
Apache 2.0
metadata
{ "repo": "solace-agent-mesh", "package": "solace-agent-mesh", "disco-role": "operating" }
description
Create, inspect, install, catalog, add, and build Solace Agent Mesh plugin packages safely.
disable-model-invocation
true

02 资源文件 3

03 SKILL.md 正文

Plugin lifecycle

Use this sub-skill when the task is to create a Solace Agent Mesh (SAM) plugin package, add plugin components into a SAM project, install plugins from local paths/wheels/Git/official names, browse the plugin catalog, or build a distributable plugin artifact.

Do not use this sub-skill for:

  • Project-level sam add agent, sam add gateway, or sam add proxy scaffolding; route to project-bootstrap.
  • Starting projects, activating plugin-backed components, submitting tasks, or checking live brokers/gateways/LLMs; route to runtime-operations.
  • CI/release automation, maintainer packaging policy, or broad repository maintenance.

Fast path

  1. Identify the lifecycle operation:

    • New reusable plugin package: sam plugin create.
    • Add one plugin instance to an existing SAM project: sam plugin add.
    • Install/verify a plugin package without creating a component config: sam plugin install.
    • Browse registry/catalog UI: sam plugin catalog.
    • Build wheel/sdist artifacts: sam plugin build.
  2. Open references/plugin-workflows.md for command forms, source choices, target files, metadata rules, and side effects.

  3. Before invoking an install/add/build operation, inspect the candidate plugin tree safely:

    python sub-skills/plugin-lifecycle/scripts/inspect_plugin.py path/to/plugin --component-name my-component --project-dir path/to/sam-project
    

    Add --json for machine-readable output, --strict to fail on warnings, or --self-test to exercise the helper on an embedded tiny fixture.

  4. If a command fails or would overwrite files, use references/troubleshooting.md before retrying.

Safety boundaries

  • sam plugin install and sam plugin add can run a package-manager command and can modify the active Python environment. Use an isolated environment when possible and prefer an explicit installer such as --install-command "uv pip install {package}" or a preinstalled plugin module.
  • sam plugin add writes component YAML under configs/agents, configs/gateways, configs/workflows, or configs/plugins and overwrites an existing target file without a merge step. Always inspect the target path first.
  • sam plugin catalog starts a local web server, opens a browser when possible, contacts/clones registries, and writes catalog cache/registry state under the SAM CLI home directory. Do not run it as a dry validation step.
  • sam plugin build runs python -m build in the plugin directory and creates/updates dist/ artifacts. Build backends may use isolated build environments.
  • This sub-skill does not run native repo tests, live catalog servers, brokers, LLM calls, gateway checks, or evaluation tasks.

Minimal checklist

  • Plugin name is intentional after SAM normalization: display/name input -> kebab directory, snake Python module, Pascal display strings.
  • pyproject.toml exists and has [project].name plus [tool.<project_name_with_underscores>.metadata].type set to agent, gateway, tool, workflow, or custom.
  • config.yaml exists at plugin package root and uses __COMPONENT_*__ placeholders where component-specific names should be substituted by sam plugin add.
  • Generated source exists for the plugin type: agent/tool usually include tools.py; gateway includes app.py and component.py; workflow is YAML-first with only package boilerplate; custom includes custom Python skeleton.
  • Build metadata includes package files and force-includes config.yaml, README.md, and pyproject.toml into the installed module so sam plugin add can read them after installation.
  • Install source is unambiguous: already installed module, local directory, wheel, Git URL, git+...#subdirectory=..., or official plugin name.
  • Existing project component files are backed up or intentionally replaced before running sam plugin add.

Provenance distilled

This sub-skill was distilled from SAM plugin documentation, plugin CLI command implementations, plugin templates, plugin catalog/official registry helpers, and plugin command unit tests. Runtime guidance is self-contained and does not require opening the source repository evidence.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351