sn-planning
Skill8 starsChanged 7 months agoMITArchived repositoryView on GitHub
Convert technical specifications into step-by-step implementation plans for SignNow MCP Server. Use when creating an execution plan, breaking down a spec into tasks, or organizing implementation work into phases. Covers phase structure, layer tags, constraint checks, and the plan output format. Do NOT use for writing specs, implementing code, or writing tests.
In this repository8 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- git-commitSkill
- sn-architectureSkill
- sn-implementationSkill
- sn-spec-writingSkill
- sn-testingSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 47
Write to .plans/Plan-{TASK_NAME}.md.
- Distinctive: line 62
Constraint check: Only decision-relevant fields.
- Very specific: line 78
Constraint check: No network calls. All mocked.
- Distinctive: line 81
Update README.md if new tools added.
- Very specific: line 83
Run pytest, ruff check src/, ruff format --check src/.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Implementation Planning
Objective
Convert a Technical Specification into a rigorous, step-by-step Implementation Plan. Define WHAT needs to be done, WHERE (exact file paths), and in WHAT ORDER — so a developer agent can execute each step linearly without ambiguity.
Output Rules
| Do | Don't |
| Write function signatures (name, params, return type) | Write function bodies or class implementations |
| Describe logic in prose/pseudo-code | Write actual code blocks |
| Use explicit file paths relative to project root | Leave locations ambiguous |
Use Markdown checkboxes - [ ] for all steps | Use numbered lists without checkboxes |
| Prefix each step with a layer tag | Mix layers within a single step |
| Describe test intent and coverage in prose | Write test code |
Layer Tags
Prefix every step:
| Tag | Layer | Location |
[API Model] | Pydantic API models | signnow_client/models/ |
[API Client] | HTTP client methods | signnow_client/client_*.py |
[Tool Model] | Curated response DTOs | sn_mcp_server/tools/models.py |
[Tool Logic] | Business logic | sn_mcp_server/tools/<feature>.py |
[Tool Orchestrator] | MCP tool wiring | sn_mcp_server/tools/signnow.py |
[Registration] | Tool registration | sn_mcp_server/tools/__init__.py |
[Config] | Configuration | */config.py |
[Auth] | Authentication | sn_mcp_server/auth.py, token_provider.py |
[Transport] | CLI/HTTP app | sn_mcp_server/cli.py, app.py |
[Tests] | Unit tests | tests/unit/ |
[Docs] | Documentation | README.md |
Plan Structure
Write to .plans/Plan-{TASK_NAME}.md. Group steps into phases following the dependency graph — foundational layers first, dependent layers last.
Phase 1: API Models
Pure data definitions. Zero dependencies. Represent SignNow API payloads.
- Each step: file path, model purpose, fields.
- Constraint check: Models import ONLY from
pydantic.
Phase 2: API Client Methods
HTTP communication. Depends only on Phase 1.
- Each step: file path, method signature, brief logic description.
- Isolation rule: No
sn_mcp_server.*imports. Errors viasignnow_client/exceptions.py.
Phase 3: Tool Response Models
Token-efficient DTOs for agents. Depends on pydantic only.
- Each step: file path, fields included/excluded with reasoning.
- Constraint check: Only decision-relevant fields. No raw API passthrough.
Phase 4: Tool Business Logic
Stateless transformation. Depends on Phase 2 + Phase 3.
- Each step: file path, function signature, numbered pseudo-code.
- Isolation rule: Transport-agnostic. No Starlette. No cross-tool imports. Token = parameter.
Phase 5: Tool Orchestrator & Registration
Thin wiring. Depends on Phase 4.
- Each step: tool function, params, docstring intent, delegation target.
- Registration in
tools/__init__.pyif needed. - Constraint check: Thin wrappers only — resolve token → call logic → return.
Phase 6: Tests
Unit tests for Phase 4. Mock SignNowAPIClient.
- Each step: test file, coverage intent, individual test names + assertions in prose.
- Constraint check: No network calls. All mocked. Runnable with
pytest.
Phase 7: Documentation & Verification
- Update
README.mdif new tools added. - Verify docstrings.
- Run
pytest,ruff check src/,ruff format --check src/. - Manual verification instructions.
Step Granularity
Each step must be atomic: "Add model X to file Y", "Add method Z to class W". A developer should complete one step, commit, and move to the next.
Philosophy Checklist
Before finalizing, verify:
- Every response model carries ONLY minimum data? (Token Efficiency)
- Tool count minimized? Could an existing tool handle this? (Tool Minimization)
- Every business logic function testable with mocked client? (Testability)
- All error messages specific — operation, entity ID, cause? (Specific Errors)
- Zero state, zero caching, zero infrastructure coupling? (Stateless)
- Feature actually needed now? (YAGNI)
signnow_client/has zero imports fromsn_mcp_server/? (No Upward Imports)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

