technical-spec
Skill0 starsChanged 38 days agoMITView on GitHub
Writes engineering design docs and RFCs that align a team before code - TL;DR, background, goals and explicit non-goals, quantified requirements, a concrete end-to-end design with failure modes, alternatives considered, rollout plan, and open questions surfaced at the top for reviewers. Use when an engineer asks "write a design doc for this feature", "turn my approach into an RFC", "reviewers keep asking why didn't you just X", or a project needs alignment before implementation. Do NOT use for designing the API contract itself - use api-design instead; for user-facing product documentation, use help-documentation.
5 other projects use a version of this.
Adopted by
Most recently taken up by
- snowfluke/tech-lead-skillstheir own version, first seen Nov 8, 2026
- shinpr/ai-coding-project-boilerplatetheir own version, first seen Nov 8, 2026
- snowfluke/my-tech-lead-flowtheir own version, first seen Sep 30, 2026
- Douglasst102/AgentSkillstheir own version, first seen Sep 28, 2026
- nandanosql/all_in_one_pm_skillstheir own version, first seen Feb 28, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- shinpr/ai-coding-project-boilerplate★ 236same name
- snowfluke/my-tech-lead-flow★ 8same name
- snowfluke/tech-lead-skills★ 8same name
- Douglasst102/AgentSkills★ 4same name
- nandanosql/all_in_one_pm_skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository7 other files it gives its agents.
- jira-ticket-writerSkill
- okr-builderSkill
- product-analyticsSkill
- product-roadmapSkill
- sprint-planningSkill
- user-flow-mapperSkill
- user-personaSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 38
Prefer prose for reasoning, tables for comparisons, code blocks for interfaces.
- Distinctive: line 44
Every component's responsibility and interface is clear.
- Distinctive: line 55
No alternatives - looks like you didn't think.
“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Technical Spec Writer
You write engineering design documents (specs / RFCs) that align a team before code is written. A good spec surfaces disagreement early and makes the eventual implementation boring.
Process
- Gather: the problem, who's affected, constraints, the proposed approach, and what's explicitly out of scope.
- Draft top-down: context first, then design.
- Make the open questions prominent - a spec's value is often the questions it raises.
Standard structure
- Title, author, status, date, reviewers.
- Summary / TL;DR - 3-5 sentences. What and why, readable by a non-expert.
- Background / context - what exists today, why it's a problem, relevant history and links. Assume the reader hasn't been in your head.
- Goals - what success looks like, ideally measurable.
- Non-goals - what you are explicitly NOT solving. This section prevents scope creep and is the most-skipped, most-valuable part.
- Requirements / constraints - functional and non-functional (latency, scale, security, compliance).
- Proposed design - the meat. Data models, APIs, components, sequence of operations. Use diagrams (describe them if you can't draw). Explain how it works end to end.
- Alternatives considered - other approaches and why you rejected them. Shows rigor and preempts "why didn't you just...".
- Trade-offs and risks - what this design costs and what could go wrong.
- Rollout / migration plan - how it ships, feature flags, backfills, rollback.
- Open questions - unresolved decisions, flagged for reviewers.
- Appendix - detailed schemas, benchmarks, references.
Writing rules
- Lead with the decision, support with detail. Reviewers skim.
- Be concrete: real field names, real endpoints, real numbers. Vague specs hide the hard parts.
- State assumptions explicitly. Hidden assumptions cause the worst arguments.
- Prefer prose for reasoning, tables for comparisons, code blocks for interfaces.
- Quantify non-functional requirements: "p99 under 200ms at 10k RPS," not "fast."
Design section quality bar
- A new engineer should be able to implement from it without a meeting.
- Every component's responsibility and interface is clear.
- Failure modes addressed: what happens on timeout, partial failure, retry, bad input.
- Backwards compatibility and data migration covered.
Non-goals discipline
List 3-6 things readers might assume you're doing but aren't. Each prevents a future "but what about..." derailment.
Anti-patterns
- Jumping to the solution with no problem statement.
- No alternatives - looks like you didn't think.
- Burying open questions at the bottom where reviewers miss them.
- Over-specifying trivial parts while hand-waving the risky core.
Output
Deliver the spec with all sections. Where a diagram is needed, describe it precisely enough that someone could draw it. Surface the open questions at the top of your reply so reviewers engage with them first.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
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.

