knowns-dev/knowns/kn-go
Claude Code and agent skill by knowns-dev/knowns
- community
Checking that the SKILL.md file is still in the repository.
Use only when the user explicitly wants the legacy no-review-gates pipeline for an approved spec; prefer kn-flow for normal spec orchestration
Install
npx vskill@latest install knowns-dev/knowns/kn-goSKILL.md
Go Mode — Legacy Full Pipeline
Announce: "Using kn-go for spec [name]."
Core principle: APPROVED SPEC → TASKS → PLAN → IMPLEMENT → VERIFY → COMMIT GATE.
kn-go is a compatibility workflow. Use it only when the user explicitly requests the legacy no-review-gates pipeline. For normal approved-spec execution, delegation, parallel safety checks, or integrated review, use /kn-flow @doc/<spec-path>.
Inputs
- Approved spec ref
- Optional
--dry-runto validate and preview tasks without mutating project state
Stop if the spec is draft, has no ACs, has blocking open questions, or the requested ref cannot be resolved.
Phase 1: Validate and Discover
- Read the spec and its complete
Locked Decisionssection. - Validate the spec.
- Resolve existing linked tasks before generating anything.
- Reuse existing tasks; skip done tasks and continue todo/in-progress work in dependency order.
- Retrieve only relevant accepted/current System Decisions with bounded filters.
With --dry-run, stop after presenting validation plus the task create/reuse preview.
Phase 2: Create Missing Tasks
Use the same task-shaping rules as /kn-plan --from @doc/<spec-path>, but skip its approval gate because the user explicitly selected go mode:
- clean titles and stable
order; no synthetic[<slug>-NN]bracket prefixes - the spec's optional
Task Generation→Task Prefixpassed asprefixon each
create call, or omitted when blank to use the project default
speclink andfulfillsmapping- outcome-oriented task ACs
- concise descriptions; implementation detail belongs in task plans
- no duplicate task scope
Report the number of reused and created tasks before implementation begins.
Phase 3: Execute Sequentially
For each pending task:
- Take ownership and start its timer.
- Follow refs and gather bounded project context.
- Read every applicable Spec Decision and retrieve relevant current System Decisions.
- Draft and save an executable plan without a per-task approval gate.
- Implement the plan; check ACs only after outcomes are complete.
- Run targeted tests, lint/build checks, and task validation.
- Record
Spec Decision Compliancefor every D-ID. - Record exactly one
System Decision Impactbranch:
none — <reason>, or- persisted draft
candidate @decision/<id> (added|changed|removed) — <summary>
- Append concise notes, stop the timer, and mark the task done.
Never create Memory category decision, duplicate Spec Decisions into the System Decision ledger, or auto-accept a Decision candidate.
If a task fails verification, keep it in-progress or blocked with the exact reason. Do not check incomplete ACs. Skip dependents whose prerequisites are blocked.
Phase 4: Integrated Verification
After runnable tasks finish:
- run SDD validation
- confirm all linked tasks are done or explicitly blocked
- confirm all Spec Decision compliance and System Decision impact markers
- run project-level test/build/lint commands appropriate to the repository
- report coverage gaps and failures exactly
Do not claim completion while validation or required verification fails.
Phase 5: Commit Gate
Commit remains the one mandatory user gate in go mode:
- Review the actual staged scope and exclude unrelated changes.
- Propose one conventional commit message.
- Ask for explicit approval before committing.
- Do not append generator or AI attribution unless explicitly requested or required by project conventions.
Context and Re-run Behavior
When context becomes constrained, finish the current safe checkpoint, report completed/blocked/remaining tasks, and instruct the user to invoke the same spec again. Re-runs must resolve existing tasks, skip done work, and continue without duplicating tasks or ACs.
Shared Output Contract
Return information in this order:
- Goal/result — what the pipeline completed, partially completed, or blocked.
- Key details — tasks reused/created/completed, blockers, SDD coverage, Decision compliance, and verification status.
- Next action — commit approval or the exact unblock/resume command.
Checklist
- [ ] User explicitly requested legacy go mode
- [ ] Spec is approved and valid
- [ ] Existing tasks resolved before creating missing tasks
- [ ] Task ACs are outcome-oriented and checked only after completion
- [ ] Each task was planned, verified, validated, and lifecycle-complete
- [ ] Decision compliance and impact markers are complete
- [ ] Integrated SDD and project verification passed or failures were reported
- [ ] Commit waits for explicit approval
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckinginternal/instructions/skills/kn-go/SKILL.md
Popularity
- Repository stars
- 213
- Repository forks
- 44
- 7-day trend score
- 27
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/knowns-dev/knowns/kn-go)