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
- 参考文档references/plugin-workflows.md11.9 KB
- 参考文档references/troubleshooting.md9.5 KB
- 脚本scripts/inspect_plugin.py22.3 KB
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, orsam add proxyscaffolding; route toproject-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
-
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.
- New reusable plugin package:
-
Open references/plugin-workflows.md for command forms, source choices, target files, metadata rules, and side effects.
-
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-projectAdd
--jsonfor machine-readable output,--strictto fail on warnings, or--self-testto exercise the helper on an embedded tiny fixture. -
If a command fails or would overwrite files, use references/troubleshooting.md before retrying.
Safety boundaries
sam plugin installandsam plugin addcan 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 addwrites component YAML underconfigs/agents,configs/gateways,configs/workflows, orconfigs/pluginsand overwrites an existing target file without a merge step. Always inspect the target path first.sam plugin catalogstarts 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 buildrunspython -m buildin the plugin directory and creates/updatesdist/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.tomlexists and has[project].nameplus[tool.<project_name_with_underscores>.metadata].typeset toagent,gateway,tool,workflow, orcustom.config.yamlexists at plugin package root and uses__COMPONENT_*__placeholders where component-specific names should be substituted bysam plugin add.- Generated source exists for the plugin type: agent/tool usually include
tools.py; gateway includesapp.pyandcomponent.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, andpyproject.tomlinto the installed module sosam plugin addcan 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
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.
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.
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.
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.
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.
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).