Skip to content

LiorCohen/sdd / plugin/core/skills/techpacks

techpacks

Verified

Single gateway for all core↔tech-pack interactions. Reads manifests, resolves paths, loads skills/agents, routes commands.

Techpacks Gateway

Single entry point for all core↔tech-pack interactions. No other core file reads techpack.yaml directly, resolves tech pack paths, or loads tech pack skills on its own.

Operations

Eight typed operations — see resources/operations.md for detailed implementations.

OperationPurposeReturns
readManifest(namespace)Parse and return validated techpack.yamlTyped manifest object
resolvePath(namespace, path)Resolve a relative path from manifest to absolute pathAbsolute file path
loadSkill(namespace, skillPath)Load a tech pack skill into contextSkill content
loadAgent(namespace, agentRef)Read agent file, parse frontmatter, resolve skills, spawn as Task subagentTask subagent with composed prompt
routeCommand(namespace, command, args)Invoke the command router skill with structured contextCommand result
routeSkills(namespace, phase, componentType?, agent?)Invoke the skills router to get relevant skills for a phaseSkills loaded into context
listComponents(namespace)Read component types from manifestComponent type list with metadata
dependencyOrder(namespace)Build topological order from dependency graphOrdered component type list

Invocation Pattern

Core skills invoke techpacks operations using this pattern:

Invoke techpacks.readManifest for the active tech pack namespace.
Read the `components` section to find the agent assigned to a given component type.
Invoke techpacks.routeSkills with:
  namespace: <active-namespace>
  phase: plan-generation
  component_type: <type>
Invoke techpacks.loadAgent with:
  namespace: <active-namespace>
  agentRef: { name: "my-agent", path: "agents/my-agent.md" }
  skills: ["coding-standards", "testing-standards"]

Active Tech Packs

Read sdd/sdd-settings.yaml (or .sdd/sdd-settings.yaml) to find active tech packs:

techpacks:
  <namespace>:
    name: <tech-pack-name>
    namespace: <namespace>
    version: "1.0.0"
    mode: internal           # or external, git
    path: <tech-pack-directory>

For internal tech packs, resolve path relative to the plugin root directory. For git-mode tech packs, resolve install_path relative to the project root.

# Git-mode example:
techpacks:
  my-pack:
    name: my-pack
    namespace: my-pack
    version: "1.0.0"
    mode: git
    repo: https://github.com/org/my-pack.git
    ref: v1.0.0
    install_path: sdd/.techpacks/my-pack/techpack

Enforcement Rules

  1. Single gateway. All core→tech-pack interactions flow through this skill. No other core file reads techpack.yaml directly.
  2. No direct reads. Core skills and system commands never read tech pack skill files or agent files directly. They always go through techpacks.
  3. Prompt isolation. Agent markdown bodies and resolved skill contents NEVER appear in the main conversation context. Use system-run.sh agent frontmatter to extract only structured metadata. Spawn agents as Task subagents with their own context.
  4. Structured logging. Every techpacks operation logs via system-run.sh log write --level info --source techpacks.<operation> --message "<description>".
  5. Fail loudly. If a manifest is missing, invalid, or a referenced path doesn't exist, return a clear error. Never silently fall back to hardcoded defaults.

System CLI Commands

The techpacks gateway uses these system CLI commands:

CommandPurpose
system-run.sh tech-pack validate --path <dir>Validate a manifest
system-run.sh tech-pack info --namespace <ns>Read manifest data
system-run.sh tech-pack listList installed tech packs
system-run.sh agent frontmatter --path <file>Extract agent metadata (prompt isolation)
system-run.sh log write --level <l> --source <s> --message <m>Structured logging
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall