niksavis
- 82 skills
- 0 followers
- 2 repo stars
- last week last updated
- ▌ Conventional Commits · niksavis bundleWrites a commit subject that passes the hooks: type, scope, breaking-change marker, record id. Use for a commit message or after a hook refused one.
- ▌ Validate As Consumer · niksavisRuns a verified change the way a consumer does, against its requirement. Use at VALIDATE or before a README claims a capability.
- ▌ Decompose Plan · niksavis bundleSplits an epic or feature into child tasks a build can check: EARS criteria, scope, dependencies and a demonstration command. Use to split work into tasks, or when the plan gate refuses a child.
- ▌ Harness Client · niksavis bundleAttaches to a running supervisor: shows what the factory does and records answers to its decisions. Use while a supervisor runs or a lane waits on a human judgment.
- ▌ Session Finish · niksavis bundleSummarizes what changed and what is still open. Use to wrap up a session or run a retro. Write only new recovery facts.
- ▌ Tier Injection · niksavis bundleChooses the model for an owner, a delegated agent, an advisor or an orchestrator, and installs the tier kit that pins it. Use when deciding which model or tier a task or subagent gets, or when a subagent ignores its tier.
- ▌ Interface Facts · niksavis bundleEstablishes a third-party CLI flag, API field, model, price or limit from the tool and live vendor docs. Use before code depends on it.
- ▌ Release Process · niksavis bundleCuts a release with basicly release, pushes it and confirms it published. Use when cutting, tagging or checking a release.
- ▌ Repair In Place · niksavisFixes a named defect in the lane's own worktree from its findings, with no new plan. Use at REPAIR after verify or validate failed.
- ▌ Test Discipline · niksavis bundleWrites isolated, order-independent tests that assert what the caller sees, not private helpers. Use for tests with shared fixtures, leaked state, run order or platform differences.
- ▌ Catalog Authoring · niksavis bundleAuthors catalog sources in YAML (always-on fragments, output styles and the skill.yaml format) and projects them. Use before editing a generated file such as SKILL.md, CLAUDE.md or AGENTS.md, or when adding a fragment or output style.
- ▌ Worktree Isolation · niksavis bundleIsolates work in a sibling git worktree, then merges and cleans up. Use to keep a change out of the main checkout or when parallel tracks collide.
- ▌ Node · niksavis bundleRuns Node and npm for the markdownlint hook and other tools. Use when npm or npx fails or resolves the wrong node binary, often on WSL.
- ▌ Python · niksavis bundleWrites typed Python with pathlib and Windows-safe subprocesses, and judges if an abstraction earns its keep. Use for .py edits, an except clause, a size gate or a silenced warning.
- ▌ Root Cause · niksavis bundleFinds the cause of a failure, a bug that came back or a test that fails intermittently, link by link with evidence, to a control this repository owns. Use when asked why something broke, keeps breaking or fails, and the cause is unclear.
- ▌ Harness Loop · niksavis bundleDrives tracked work through the basicly loop, intake to ship. Use to start or resume work, or to move an issue past a checkpoint or gate.
- ▌ Work Tracker · niksavis bundleReads, files, claims and closes records in the ledger tracker. Use to plan work, check what is ready, file a bug, or claim a record before work starts.
- ▌ Falsify First · niksavis bundleTries to break a claim, invariant or measurement with a concrete counterexample. Use before it enters a plan, a design or a gate, or before a high-risk number, root cause or coverage claim enters a report.
- ▌ Retention Probe · niksavisTests if this session still holds the always-on instruction file. Use when a repo rule seems missing, in a new subagent or after compaction.
- ▌ Best Practices Audit · niksavis bundleAudits skills, instructions, hooks, agents and permissions against the newest vendor docs. Use after a Claude Code release, or when asked if the harness still follows Anthropic guidance.
- ▌ Swarm · niksavis bundleRuns several fresh agents on one artifact, one lens each, and merges their evidence into a verdict table; in teardown mode it also verifies, answers and fixes every finding. Use when asked for an agent swarm, several reviewers, a multi-agent review, or to tear down or find everything wrong with a diff, plan, design or document.
- ▌ Council · niksavis bundleDecides one closed question by independent votes of fresh agents with different stances, and reports the majority with the dissent. Use when asked to let a council, a panel or several agents vote on a choice.
- ▌ CLI Tools · niksavis bundlePicks the shell tool and its flags: rg, fd, jq, yq, curl, shellcheck, sd, ast-grep, typos, git, uv or tmux. Use to find files that mention a word, list files, read a JSON or YAML field, call an HTTP endpoint and show the response, or replace text.
- ▌ Hill Climb · niksavis bundleImproves a plan, design, code or skill in graded rounds: fresh reviewers grade fixed properties 0 to 10, the author revises, and a script decides when to stop. Use when asked to hill-climb, iterate to a better version or improve something in rounds.
- ▌ Find Skills · niksavis bundleSearches the catalog, then skills.sh, for an existing community skill out there, and installs one only after the user agrees. Use when asking: is there a skill for this?
- ▌ No Comments · niksavisKeeps prose comments and docstrings out of code: where a fact goes instead and which tool directives stay. Use when writing a comment or docstring, or when the no-comments gate refuses a commit.
- ▌ Plain English · niksavisWrites plain prose for readers of any language: READMEs, release notes, design docs, records. Use when writing text for a person.
- ▌ Skill Creator · niksavis bundleWrites a new skill or improves one with evals, a benchmark, graded versions and description tuning. Use when adding a new skill to the catalog, turning a workflow into a skill, comparing runs with and without a skill, or when a skill never gets picked up.
- ▌ Python Guidelines 2 · niksavisMake the design calls no linter can check — where an oversized file splits, whether a name or docstring carries meaning, whether an abstraction earns its keep, when a suppression is legitimate, and how to satisfy a size or complexity ratchet without gaming it. Use when a size or complexity gate has just failed, before adding a noqa or nosec, when deciding what to raise and what to catch, or when shared state is reached from more than one concurrent lane.
- ▌ Work Tracker 2 · niksavisUse the owned work tracker - the append-only event ledger under .basicly/ledger/ - as the primary task/issue tracker for this repo, reading it through the kit CLI and writing it through the engine seam, and know what it refuses. Trigger when planning work, creating or claiming an issue, checking what is ready to work on, counting or querying issues in bulk, or preparing a commit that must reference a tracker issue id.
- ▌ Falsify First 2 · niksavisAttempt to break a claim - an invariant, a design premise, a measurement - with a concrete counterexample search before defending or adopting it, and read each kill for the precondition that carried the weight. Use before an invariant enters a plan, a design note or a gate, before a measured number becomes a claim, and whenever a candidate rule survived only because nobody attacked it.
- ▌ Decompose Plan 2 · niksavisCut a unit of work into children the plan gate will accept — testable acceptance criteria in EARS, disjoint or declared scope globs, an acyclic dependency graph, a token budget, an integrity level, and the command that demonstrates each child end to end. Use at DECOMPOSE, when a plan gate has just refused a child, or when a child turns out to have no consumer-visible behaviour to check.
- ▌ Harness Client 2 · niksavisAttach to a running basicly supervisor as a second session — observe live status, present its pending decisions to a human conversationally, and record the answers. Use when a supervisor is already running (or may be) and you are not the one driving it, to check what the factory is doing, unblock a lane that waits on a judgment, or answer a queued decision.
- ▌ Session Finish 2 · niksavisClose out a working session with a usage-statistics report, a self-improvement retro, and a pickup-clean handoff summary. Use when the user says the session is done ("wrap up", "finish the session", "close out"), before ending a long autonomous run, or whenever a summary of what changed and what the agent actually used is wanted.
- ▌ Tier Injection 2 · niksavisInstall the portable tier injection kit so a subagent spawns on the model its declared tier resolves to, instead of the host default. Use when setting up tier injection in this or another repository, when a subagent ignores the tier its definition declares, or when deciding whether a host can pin a spawn's model at all.
- ▌ Interface Facts 2 · niksavisEstablish a fact about an external interface - a CLI flag, an API field, a model id, a price, a limit, a version - by fetching the vendor's current documentation instead of recalling it. Use before writing code, a design note, or any claim that depends on how a third-party tool behaves, and whenever a repo document already asserts such a behaviour.
- ▌ Release Process 2 · niksavisCut a release of this repository with `basicly release`, then do the three steps that command deliberately leaves to a human - deciding the version, pushing, and replacing the release page with the highlights. Use when asked to cut a release, tag a version, prepare release notes, or check whether a release published.
- ▌ Repair In Place 2 · niksavisFix a named defect in the lane's own worktree, briefed with the actual findings and without re-planning or widening scope. Use at REPAIR after verify or validate failed, when a landing bounced, or whenever the temptation is to re-read the requirement and start again.
- ▌ Test Discipline 2 · niksavisWrite isolated order-independent automated tests that assert on observable behavior rather than private internals. Use when writing reviewing or debugging any test (unit integration or end-to-end) in any language especially when tests share fixtures touch global or filesystem state flake depending on run order or reach into implementation details.
- ▌ Catalog Authoring 2 · niksavisAuthor and improve basicly catalog sources — skills and fragments — in their YAML source format (never a discoverable .md), then project and verify them. Use when adding or editing a skill or fragment, building a catalog, or deciding where guidance should live (always-on fragment vs on-demand skill).
- ▌ Worktree Isolation 2 · niksavisIsolate non-trivial work in a sibling git worktree using `basicly worktree`, covering sibling placement on a harness branch, dependency + git-hook provisioning, and safe cleanup. Use when starting a unit of work that should not touch the main checkout, when parallel tracks would collide, or when deciding whether a change needs its own worktree.
- ▌ Conventional Commits 2 · niksavisConstruct a valid Conventional Commits message for this repo before running `git commit`, covering type/scope, the "!" breaking-change marker, description rules, and the required trailing tracker record id. Use whenever writing or reviewing a commit message, or when a commit is rejected by the commit-msg/tracker-commit-msg hooks.
- ▌ Validate As Consumer 2 · niksavisExercise a verified change the way a consumer would — in the operational environment, against the requirement that asked for it — instead of re-running the gate suite that already passed. Use at VALIDATE, before claiming a capability on a README or release note, or whenever "the tests pass" is standing in for "the feature works".
- ▌ Node 2 · niksavis bundleUse Node and npm in this repo for the markdownlint git hook and other node tooling. Use when running npm or npx committing or pushing from a script or background job on WSL or debugging a node-based hook that resolves the wrong node binary.
- ▌ Python 2 · niksavis bundleWrite and edit Python for this repo — type hints pathlib and cross-platform subprocess and shell-out. Use when creating or changing .py files wiring up a subprocess call chasing a test that passes on POSIX but fails only on Windows CI (a WinError 2 or a mangled backslash path) or second-guessing syntax that looks wrong for an older Python (an unparenthesized multi-exception except clause).
- ▌ Root Cause 2 · niksavisEstablish why something actually happened by iterating why until the answer stops changing, with each link citing an observation - and know the two ways the method lies. Use before filing a bead off a failure, before proposing a rule or a gate to prevent a recurrence, when a fix addresses a symptom, or when a retro asks for a cause.
- ▌ Harness Loop 2 · niksavisDrive a unit of work through the basicly harness loop end-to-end (intake → classify → decompose → build → verify → validate → ship → done) with `basicly loop` over the owned tracker ledger, agent-agnostic across Claude/Codex/Copilot. Teardown is folded into the ship advance and the retro is a tracker comment; neither is a phase. Use when starting or resuming non-trivial development in a harness-enabled repo, when deciding what phase a tracked issue is in, or when coordinating the checkpoints, gates, and bounded rework the loop enforces.
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌ Python Guidelines · niksavisMake the design calls no linter can check — where an oversized file splits, whether a name or docstring carries meaning, whether an abstraction earns its keep, when a suppression is legitimate, and how to satisfy a size or complexity ratchet without gaming it. Use when a size or complexity gate has just failed, before adding a noqa or nosec, when deciding what to raise and what to catch, or when shared state is reached from more than one concurrent lane.