kamiflow-core
Route daily Codex work through a clarity-first, client-repo-first workflow with a repo contract in `AGENTS.md`, human project memory in `.local/project.md`, and helper-backed task plans under `.local/plans/`. Use when Codex needs to infer the lightest safe lane from prompts like brainstorm, idea, plan, implement, fix, review, verify, or investigate; shape unclear work before implementation; write a decision-complete plan; implement in scoped slices; verify with evidence; recover plan state inside any client repo; or honor an optional SemVer release policy in `AGENTS.md` for opted-in Node/npm repos. Treat the kamiflow-core source repo as the source-repo exception.
94
92%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Advisory
Suggest reviewing before use
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong, detailed description that clearly articulates a specific workflow methodology with concrete actions, explicit trigger conditions, and distinctive terminology. It covers both what the skill does and when to use it comprehensively. The main weakness is that it is quite dense and jargon-heavy (e.g., 'lightest safe lane', 'clarity-first', 'helper-backed task plans'), which could make it harder to parse quickly, though the specificity and completeness are excellent.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: routing work through a clarity-first workflow, maintaining a repo contract in AGENTS.md, managing project memory in .local/project.md, creating helper-backed task plans, writing decision-complete plans, implementing in scoped slices, verifying with evidence, recovering plan state, and honoring SemVer release policy. | 3 / 3 |
Completeness | Clearly answers both 'what' (clarity-first workflow with repo contracts, project memory, task plans, scoped implementation, verification) and 'when' with an explicit 'Use when...' clause listing specific trigger scenarios like inferring lanes from prompts, shaping unclear work, writing plans, implementing, verifying, and recovering plan state. | 3 / 3 |
Trigger Term Quality | Includes natural trigger terms users would say: 'brainstorm', 'idea', 'plan', 'implement', 'fix', 'review', 'verify', 'investigate'. Also includes contextual terms like 'Codex', 'AGENTS.md', 'SemVer', 'Node/npm'. Good coverage of natural workflow-related language. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with a clear niche: it targets a specific 'kamiflow' workflow methodology for Codex with very particular file paths (AGENTS.md, .local/project.md, .local/plans/), specific concepts (clarity-first, lightest safe lane, decision-complete plans), and even names the source repo exception. Unlikely to conflict with other skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured, highly actionable skill that provides clear routing logic, concrete commands, explicit validation checkpoints, and good progressive disclosure through well-signaled references. Its main weakness is moderate verbosity—some rules are repeated across sections (ready-check hard stop, client-repo-first emphasis), and the Design Rubric section, while valuable for skill maintenance, adds token cost that may not serve everyday Codex operation. Overall, it's a strong skill that effectively guides complex multi-route workflow decisions.
Suggestions
Consolidate repeated rules (e.g., ready-check hard stop appears in Quick Start, Core Workflow, and Boundaries) into a single authoritative location to reduce token cost.
Consider moving the Design Rubric section to a separate reference file (e.g., references/design-rubric.md) since it serves maintainers rather than daily Codex operation.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is fairly well-organized and avoids explaining basic concepts, but it's quite long with some redundancy (e.g., the ready-check hard stop rule is stated in Core Workflow, Plan Contract, and Boundaries). The Design Rubric section, while useful for maintainers, adds significant token cost for everyday use. Some sections could be tightened. | 2 / 3 |
Actionability | The skill provides concrete, executable commands (e.g., `node .agents/skills/kamiflow-core/scripts/ensure-plan.mjs --project .`), specific routing tables mapping prompt shapes to routes with first moves, clear output contracts (State/Doing/Next + PASS/BLOCK), and explicit decision rules like the ready-check hard stop. The guidance is specific and directly usable. | 3 / 3 |
Workflow Clarity | The workflow is clearly sequenced with explicit validation checkpoints: ready-check as a hard stop before build/fix, check-closeout for slice closing, archive only after all acceptance criteria are checked. There are clear feedback loops (failing ready-check reroutes to plan-only work), and the route quick-pick table provides unambiguous sequencing. The unknown-type matrix handles ambiguous cases. | 3 / 3 |
Progressive Disclosure | The skill is an excellent overview that delegates detailed content to well-signaled one-level-deep references (route-intent.md, command-map.md, individual route references like start.md/plan.md/build.md, and assets/*.md). Helper commands are organized into buckets with details deferred to command-map.md. Advanced helpers are clearly separated from the core daily-use path. | 3 / 3 |
Total | 11 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- kamishino/kamiflow-codex-starter
- Commit
09513b6
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.