DataArt/artisyn-implementation-harness / skills/implementation-harness
implementation-harness
FlaggedInstall the portable AI-dev harness, rebuild its operational graph standalone, harvest project skill improvements into its canonical templates, package direct canonical edits into a portable bundle, or apply canonical updates to an existing project. Use /implementation-harness launch, /implementation-harness reindex-graph, or /implementation-harness patch harvest|prepare|apply.
canonical/agents/layer/INTENT.md
View on GitHub---
skill: agents/layer
output: {{provider:agents_root}}/{{LAYER_NAME}}.{{provider:agent_extension}}
---
## Purpose
Generic specialist agent for any bounded implementation layer (web, API/BFF, data, worker, library, CLI, or infrastructure). Instantiate once per detected layer. The generated filename follows the actual layer name; the canonical source is always this profile-driven template.
## Substitutions
| Placeholder | Source | Notes |
| -------------------------------- | ---------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------ |
| `{{LAYER_NAME}}` | Bootstrap Q — layer name | Actual component name discovered from architecture/source ownership; used for agent identity and scope |
| `{{LAYER_PATH}}` | Bootstrap Q — layer path | e.g. `apps/web/`, `apps/lambda-auth/` |
| `{{LAYER_DESCRIPTION}}` | Bootstrap Q — one-line role label | e.g. "VitePress frontend specialist", "Lambda@Edge auth handler" |
| `{{LAYER_ROLE}}` | Bootstrap Q — full role paragraph | What this layer owns; what architectural role it plays |
| `{{LAYER_EXCLUDED_CONCERNS}}` | Bootstrap Q — explicit non-scope | What this agent must not touch (e.g. "CDK infrastructure, VitePress config, docs content") |
| `{{NX_APP_NAME}}` | Bootstrap Q — NX project name | e.g. `site`, `lambda-auth`, `bff`. Used in `yarn nx run {{NX_APP_NAME}}:build` |
| `{{LAYER_KEY_RULES}}` | Bootstrap Q — 2–4 enforcement rules | Layer-specific invariants (timing-safe compares, accessibility, p95 targets, etc.) |
| `{{LAYER_REVIEW_TRIGGERS}}` | Bootstrap Q — non-trivial examples | e.g. "new auth flow, header handling change, credential validation logic" |
| `{{LAYER_REVIEW_SKIP_EXAMPLES}}` | Bootstrap Q — mechanical examples | e.g. "config value, log format, minor refactor with no auth logic impact" |
| `{{GUARDED_COMMANDS}}` | Q7 — guarded commands | Commands to present as instructions, never run directly |
| `{{LAYER_NEVER_DO}}` | Bootstrap Q — additional prohibitions | Layer-specific hard rules (e.g. "Never use === for credential comparison") |
| `{{LAYER_PROFILE}}` | Derived from Q2/Q3 and source inspection | `web`, `backend`, `data`, `worker`, `library`, `cli`, `infrastructure`, or another confirmed profile |
| `{{LAYER_TOOLCHAIN}}` | Repository build/config evidence | Actual language, framework, package/build tool, compiler and check commands; never assume TypeScript, Nx, Yarn or package.json |
| `{{LAYER_PATTERNS}}` | Source/config evidence and decisions | Profile-specific rules for contracts, state/data, configuration, operations and diagnostics |
## Conditionals
- If the layer has no Nx project, TypeScript or package.json: remove those rules and use the detected toolchain and exact configured commands
- If a profile has no applicable concern (for example no persistence for a worker), omit it and record non-applicability in the launch record
- If there are no additional `{{LAYER_NEVER_DO}}` rules: leave the placeholder empty or remove the trailing block
## Bootstrap pattern
Instantiate one agent per confirmed implementation component, regardless of its
architectural category. Derive identity, owned paths, responsibilities, toolchain
and exclusions intelligently from project evidence. Profiles guide applicable rules;
they do not prescribe a fixed component list or separate canonical templates.
Record each instance in the root instruction file's Agents table with
`Component | Owned paths | Agent | Responsibilities`. Use that same map in Dev and
X-ray. In both-provider launches, keep identities and ownership consistent while
rendering provider-specific agent formats. Supply per-instance knowns when several
components share the `layer` template; never reuse one component's facts for another.
Infrastructure uses the same template with profile `infrastructure`; it receives deployment safety rules from detected command/config evidence.
## Verify checklist
- [ ] Five sections present: Role, Layer scope, What to enforce, Satellite review, What never to do
- [ ] `{{LAYER_NAME}}` resolved — no literal placeholder remaining in agent name, NX commands, or commit scopes
- [ ] `{{LAYER_PATH}}` resolved — no literal placeholder remaining in path references
- [ ] Any build command uses the actual component toolchain; no command is invented for an absent tool
- [ ] Satellite review section references `{{provider:skills_root}}/dev/patterns/satellite-review.md`
- [ ] `{{LAYER_KEY_RULES}}` expanded — at least one layer-specific enforcement rule present
- [ ] `{{LAYER_PROFILE}}` resolved to a confirmed profile and its rules are applicable
- [ ] `{{LAYER_TOOLCHAIN}}` contains only detected language/build/check facts
- [ ] No assumptions about TypeScript, Nx, Yarn, package.json or `src/` remain when absent
- [ ] No overlapping layer agent owns the same path
- [ ] `{{GUARDED_COMMANDS}}` resolved — matches Q7 guarded command list
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 471k | Updated | Claude CodeCodex | Verified | Install |
| 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: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| 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 available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |