waqas1412

waqas1412/claude-harness

Plugin Claude Code

Waqas Hameed's portable Claude Code harness: advisor agents, profile-driven PR/ticket skills, enforcement hooks, and a self-adapting project bootstrap.

waqas1412/claude-harness

Plugin Claude Code

Portable read-only advisor agents, profile-driven PR/ticket skills, enforcement hooks, and a self-adapting /harness-init that tailors any repo to a documented index pattern.

waqas1412/claude-harness

Agent

Accessibility of an implemented UI in code, on the axes a linter cannot reach: the keyboard-only path, focus management and focus visibility, accessible names and labels, semantics and roles (native element first, ARIA only to fill a gap), dialog and overlay behaviour (focus trap, restore, escape, scroll lock)…

waqas1412/claude-harness

Agent

Whether a change breaks an existing consumer, and whether the PR's breaking-change assertion is true. Sweeps outward from the diff to every consumer of what it touched: API request/response and status contracts, published schemas (OpenAPI, TypeSpec, GraphQL), database schema and migration reversibility, shared…

waqas1412/claude-harness

Agent

Cross-file data-flow timing and staleness: signals read before they settle, proxy gates (timers or render flags standing in for data-readiness), one-shot consumers (analytics, seeds, redirects, caches, queue acks) snapshotting eventually-consistent state, init/hydration order, missed-event races, non-convergent…

waqas1412/claude-harness

Agent

Parity of an implemented UI with its pinned design source of truth (Figma file or captured design exports): token identity before resolved values (token bypass, wrong-tier token, duplicated decision), layout as auto-layout semantics not coordinates (Fixed/Hug/Fill/min-max), the interaction x data x environment state…

waqas1412/claude-harness

Agent

Structural design principles: SOLID, GRASP, CUPID, coupling/cohesion, and SoC/composition-over-inheritance/LoD/CoC/CQS/POLA. Two gates: PLAN (shape a sound design) and VERIFY (audit a design or diff). Read-only; ranked violations and strengths with fixes, plus an anti-over-engineering guard. Not…

waqas1412/claude-harness

Agent

Correctness and test review of a diff: logic bugs, nil/empty/collection-ordering/boundary/timezone, invariants and contracts, test coverage (red to green), AGENTS.md / CLAUDE.md compliance. Two gates: PLAN (risk and test plan) and VERIFY (adversarial diff review). Read-only; returns findings with severity and a fix.…

waqas1412/claude-harness

Agent

External knowledge from OFFICIAL primary sources (docs, source, types, CHANGELOG, migration guides): version-correct API usage, idioms, breaking changes and deprecations for whatever libraries, frameworks, and languages this repo actually depends on. Two gates: PLAN (research best practice before building) and VERIFY…

waqas1412/claude-harness

Agent

Drafts a tracker ticket for this repository by following the /ticket skill templates (single source of truth) and reading .claude/harness/profile.md for the tracker prefix and links. Returns a ready-to-paste ticket: terse : title, INVEST, named persona, 3 to 5 testable AC, required Out-of-scope, verification gate, no…

waqas1412/claude-harness

Agent

Performance and complexity: time/space Big-O, allocations/memory, DB N+1 and indexes, concurrency, frontend render/bundle. Polyglot (e.g. backend, web, data layers). Two gates: PLAN (set a complexity budget) and VERIFY (measured optimization audit). Read-only; returns measured, behavior-preserving recommendations. Not…

waqas1412/claude-harness

Agent

Drafts a GitHub PR title and body by following the /pr skill template (single source of truth) and the repo profile. Reads the working diff, returns ready-to-submit text with no em dashes and N/A sections deleted; the main loop runs gh pr create. Not code-correctness review (use developer-reviewer); not design (use…

waqas1412/claude-harness

Agent

Reuse, duplication, and abstraction right-sizing: DRY/WET/AHA/KISS/YAGNI, what to factor out vs leave alone, guarding against over-engineering and premature abstraction. Two gates: PLAN (shape reuse) and VERIFY (audit a design or diff). Read-only; pragmatic prioritized recommendations. Not SOLID/GRASP/coupling…

waqas1412/claude-harness

Agent

Security review of a diff along the axes a static scanner cannot judge: authorization and tenant/customer scoping, authentication and token/session handling, secrets and credential exposure, sensitive data reaching logs, analytics, URLs or client bundles, injection and deserialization at real trust boundaries, and…

waqas1412/claude-harness

Agent

Implementation craft: drafting and validating idiomatic code in the repo's primary language(s), following its conventions for error handling, data access, framework wiring, and formatting. Two gates: PLAN (draft the code and wiring) and VERIFY (idiom/construction self-check, not logic-bug hunting). Read-only advisor…

waqas1412/claude-harness

Agent

Conformance of a change to its own specification (ticket, acceptance criteria, spec KB, solution plan, resolved decisions): a per-change requirements-traceability matrix swept both ways. Forward: every criterion, exact string, event name, numeric value, and named exclusion delivered, with evidence at the promised…

waqas1412/claude-harness

Agent

Macro structure: where code should live (package/file/layer), module boundaries, blast radius, structural cross-component data-flow and contracts (shape, not timing), fit to repo patterns. Two gates: PLAN (placement decision) and VERIFY (audit implementation vs intended structure). Read-only advisor. Not exact…

waqas1412/claude-harness

Agent

Micro component spec once placement is decided: exact function/query signatures, request/response shapes, algorithm/aggregation steps, exhaustive edge-case matrix. Two gates: PLAN (write the spec) and VERIFY (built component matches the spec). Read-only advisor. Not placement/boundaries (use system-architect); not…

waqas1412/claude-harness

Hook

Runs before the agent uses a tool for Bash, Write and Edit tool calls, executing block-coauthor.sh, block-pr-reviewer.sh, block-force-push.sh and block-md-emdash.sh via sh (5 commands). From waqas1412/claude-harness.

waqas1412/claude-harness

Hook

Runs after a tool call finishes for Bash tool calls, executing filter-verbose-output.py via python3. From waqas1412/claude-harness.

waqas1412/claude-harness

Skill Claude CodeCodex

Distill durable learnings from recent sessions and code-review corrections into proposed memory facts, CLAUDE.md rule lines, or skill Gotchas. Use periodically, or after a run that surfaced recurring friction, corrections you keep repeating, or a gotcha worth capturing. It clusters candidates, adversarially verifies…

waqas1412/claude-harness

Skill Claude CodeCodex

Scan the current repository and generate a tailored navigation harness (root CLAUDE.md index, read-on-demand .claude/repo-index/.md deep indexes, .claude/meta/navigation.md, and a .claude/harness/profile.md consumed by /pr and /ticket). Runs autonomously: infers stack, areas, verify commands, and conventions, then…

waqas1412/claude-harness

Skill Claude CodeCodex

Apply the review lenses yourself, in the solo main loop, with no subagent dispatch. Pick only the lenses the change actually touches, run each as a checklist against real code and real command output, and gate go/no-go on cited evidence. Use when asked to run, use, or review with "our lenses", or to plan or verify a…

waqas1412/claude-harness

Skill Claude CodeCodex

Triage unresolved review threads on an open PR and close them out. Per thread, either agree (fix scoped to THIS PR's diff, verify, amend and force-with-lease, reply, resolve) or disagree (justify with a quoted citation from the version-matched official doc, reply, resolve). Every reply is drafted for approval before…