navapbc

Use when the user wants a comprehensive multi-issue health pass — fix all bugs, resolve broken tests, clear CI failures, fix lint and type errors, restore green builds — not for a single targeted bug fix. Diagnoses and resolves every outstanding open bug ticket plus all validation failures (test, lint, type, format, infrastructure) using orchestrated sub-agents with TDD discipline, in tier order. Trigger phrases include 'debug everything', 'fix all bugs', 'fix the failing tests', 'CI is failing', 'build is broken', 'clean up the codebase', 'restore green', 'project health pass'.

Use when wrapping up a worktree session — closing tasks, finishing work, or merging the branch back to main. Closes completed tickets, sweeps orphaned epics, extracts technical learnings, files bug tickets for unfixed failures, commits remaining changes, syncs the tickets branch, merges to main via merge-to-main.sh, verifies the worktree is merged and clean for claude-safe auto-removal, and prints a task summary. Trigger phrases include 'end session', 'wrap up', 'finish', 'done for now', 'close out', 'merge to main and clean up', 'session complete', 'wrap up this branch'.

Use when the user is stuck in a merge conflict, rebase conflict, failed pull, or any git operation that left conflict markers in the working tree. Dispatches a conflict-analyzer sub-agent that classifies each conflicted hunk by confidence (auto-resolvable vs needs-human-review), auto-resolves high-confidence cases, validates the result, and rolls back if validation fails before falling through to human review. Trigger phrases include 'merge conflict', 'resolve conflicts', 'rebase conflict', 'git conflict', 'conflict markers', 'failed rebase', 'git pull conflict', 'fix the conflicts', 'help with this merge'.

Manual review escalation when CI llm-review blocks a PR on a finding that looks like a false positive. Dispatches dso:code-reviewer-standard at opus tier on the PR's diff, parses the result, and emits a force-merge clearance when zero critical/important/fragile findings remain. Use ONLY when CI llm-review has failed on a suspect FP; not for routine PR shipping. Trigger phrases include 'force merge this PR', 'the CI review is wrong', 'false positive from llm-review', 'manual review override', 'FP recovery'.

Use when the user wants to execute an epic, run a sprint, work through a planned epic's stories and tasks, or coordinate multi-agent task execution end-to-end. Routes the epic by complexity (SIMPLE → direct implementation-plan, MODERATE → lightweight preplanning, COMPLEX → full preplanning), runs an SC-coverage gate at haiku/sonnet/opus tiers to confirm story coverage of epic success criteria, plans the task graph, dispatches sub-agents in batches with file-overlap and semantic-conflict checks, runs per-task review and post-batch validation (test gate, lint, AC verification, visual verification for UI), commits/pushes results, and verifies epic completion via the dso:completion-verifier agent before close. Trigger phrases include 'work the epic', 'execute the sprint', 'run the epic', 'sprint this epic', 'work through the stories', 'implement the planned tasks', 'kick off the sprint'.

Use when breaking down an epic into user stories, story splitting, backlog grooming, defining acceptance criteria, or auditing and reconciling existing epic children before implementation. Dispatches an opus `dso:story-decomposer` sub-agent to draft prioritized vertical-slice user stories with measurable done definitions tied to epic Success Criteria (never inline), identifies dependencies, runs an adversarial red-team review pass, dispatches a UI designer for UI stories, and writes the story tickets to the tracker. Trigger phrases include 'break down this epic', 'split into stories', 'story splitting', 'backlog grooming', 'write user stories', 'define acceptance criteria', 'plan the epic', 'decompose the epic', 'reconcile epic children'.

Use when fixing a single bug, debugging an error or crash, investigating a defect, or routing a bug report through structured investigation. Classifies the bug by type and severity, runs pre-investigation gates (intent search, feature-request check), dispatches a tiered investigation sub-agent (BASIC/INTERMEDIATE/ADVANCED) to identify root cause, applies a TDD fix loop with RED-before-GREEN discipline, and runs post-fix gates before commit. Trigger phrases include 'fix this bug', 'debug this', 'investigate this issue', 'something is broken', 'there is a problem with', 'crash', 'error', 'defect', 'regression', 'bug triage'.

Generate a project-specific CLAUDE.md file from .claude/dso-config.conf and the Digital Service Orchestra template

Emit Catala

Create Sample Tests

Draft or Update Test Cases

Expand Test Coverage

Extract Test Cases from Policy Documents

Generate a Demo App (Catala-Python Backend)

Refine Ruleset Guidance for a Domain

Review Extracted Ruleset

Runs monthly LCAT check for Finance. Pulls all billable roles from Jira Staffing board (project SO) where Project Start Date falls in a given month then creates new Google Sheet report in Delivery Staffing Movements Drive folder for Finance to compare LCATs vs. Unanet. Trigger phrases: "monthly LCAT check", "run the LCAT check", "LCAT check for [month]", "run the finance report", "monthly finance check", "pull staffing for [month]", "create the finance onboarding sheet", or "finance staffing report." What it does: - Asks which month to report on and whether to notify Finance afterward - Queries Jira (project SO) for roles with Project Start Date in that month, excluding Corporate/Non-Billable - Builds CSV with columns: Project Name, Employee Full Name, Project Start Date, Summary, Issue Key, GSA LCAT, Contract LCAT, Unanet LCAT, Flag/Comment? - Creates new Google Sheet titled {Month} {Year} - Delivery Staffing Movements in Finance LCAT reports Drive folder - Sends summary to Finance Slack channel

Input policy document (links or downloads) to generate a flowchart, service blueprint, and plain list of all policy-based delivery requirements. Best used for discovery and team alignment. Examples: - "Looking at Federal Regulation X, Memo Y and State Manual Z, what are the income requirements for AK residents to be eligible for Medicaid?" - "Based on policy in the VA's M21 manual, what are the documentation requirements for a veteran with a presumptive claim who is applying for benefits?" Results should be reviewed by a human!

Use this skill to evaluate whether a skill is ready for enterprise-wide publish in the Nava Skills Marketplace. Triggers when someone shares a SKILL.md file for review, asks whether a skill is ready to publish, says "review this skill for the marketplace," or when called via API as part of the automated submission workflow. Also use when someone asks "can I publish this?" or "does this pass the rubric?" about any skill file. Runs a two-question rubric against the SKILL.md contents, cross-references against submitter-provided form data when available, and returns a structured verdict with actionable fix instructions.

End-of-week brag-doc capture for any Nava employee pursuing a level-up. Pulls Jira and Slack from the last 7 days, ranks candidate entries against the Nava competency gaps the user is targeting, and writes confirmed entries directly into a .docx brag log on the user's computer. Use when the user says "weekly brag log", "Friday brag log", "log my wins", "update my brag doc", "promotion prep", or when scheduled to run at end-of-week.

Search Nava's Project Index spaces to find relevant past performance, similar work, and project capabilities for proposal research. Use this skill whenever someone asks to find past Nava projects, look up similar work for an active opportunity, search for past performance examples, identify which projects match an RFP requirement, find project capabilities by agency or technology, filter by portfolio (FedCiv, FedHealth, VA, State), filter by archetype (Product Team, Data Modernization Team, Enterprise Operations Team, etc.), or asks any variant of "what has Nava done with X?" or "find me projects similar to Y." Triggers on natural language queries using RFP phrasing, federal evaluation language, agency names, technology keywords, dollar thresholds, portfolio names, or archetype labels. Also triggers when someone asks who to talk to about a specific project area.

Review a proposal, strategic initiative, roadmap, project plan, press release, or document through three core lenses: alignment with company goals, cross-org opportunities, and execution risks. Supports five modes: (1) Default proposal review for internal documents, (2) Press Release / External Comms mode with completeness checklist and quote quality assessment, (3) Go/No-Go Decision mode with decision readiness assessment and blocking dependency tracking, (4) Strategic Transformation mode for org restructuring and multi-workstream initiatives, (5) Training Material mode for training decks, workshops, and learning programs. Use whenever someone says "review this proposal", "what do you think of this plan", "review this doc", "give me feedback on this initiative", or references a document that needs strategic evaluation.

Run the start-of-week planning ritual for any Nava manager or leader. Produces a focused weekly goals and tasks document grounded in the user's strategic priorities, real data from Slack/Jira/Calendar, and their own input on what matters most this week. Triggers when the user says "week kickoff", "weekly planning", "start of week", "plan out the week", "let's set up my weekly goals", or similar. Invoke proactively at the start of any Monday work session where weekly goals haven't been set.

Run a fast daily briefing to orient any Nava employee to their day. Pulls calendar and Slack signals in parallel and delivers a tight, 1–2 minute summary — meeting density, prep needs, urgent Slack items. Triggers when the user says "morning briefing", "daily briefing", "good morning", "what's on my plate today", "start my day", or opens a session with a short, open-ended, greeting-like first message. Invoke proactively at the start of a workday session when no specific task is given.

Scaffolds a Nava Strata Rails application with the nava-platform CLI and navapbc/template-application-rails template, then optionally installs the Strata SDK. Use when user says build a strata app, scaffold a rails app, or apply the rails template.

Adds one Rails model via a direct single-agent test-first walkthrough — plain ActiveRecord, or a Strata SDK variant (application form, case, business process, task). Default; for the multi-agent variant use build-strata-sdk-model-workflow.

Orchestrates adding one Rails model (plain ActiveRecord, or a Strata SDK application form, case, business process, task) as a multi-agent Workflow pipeline with attribute audit and two checkpoints. Use only when opting into orchestration.

Guides deploying a Strata Rails application to Heroku, covering Node.js and Ruby buildpack ordering, git-subtree push for subdirectory repos, DATABASE_URL migration, AWS startup errors, and Cognito auth adapter bypass.

Builds views, flow, and routes for a Strata multi-page application form on top of an existing ApplicationForm model. Use after generating the model with build-strata-sdk-model — covers pages, step indicator, review, and dashboard.

Classify each failing CI test into one of four verdicts — APPLICATION_BUG, TEST_BUG, FLAKY_FAILURE, or ENVIRONMENT_ISSUE — so teams never auto-generate no-op tests for genuinely broken code, never "fix" the application when the test itself is stale, and never burn a code/test patch on an intermittent or infrastructure failure. Built on Joe's framing (is the test wrong or the code wrong?) and aligned to the industry-standard four-way failure taxonomy used by tools like ContextQA and FixSense. For each failing test the skill emits a verdict plus a short rationale and a failure category (visual drift, behavioral drift, E2E form-flow drift). Produces a human-readable terminal report and a machine-readable JSON block that the dispatcher uses to post a single PR comment requesting a mandatory 👍/👎 reaction. Use this skill whenever a CI test run fails and you want to know whether the test, the code, the test's stability, or the environment is at fault.

Perform a comprehensive pre-commit security review on uncommitted changes. Covers all areas of the built-in /security-review, OWASP Top 10 (where applicable), and mandatorily detects secrets, PII, and PHI in staged changes. Use whenever a security review is requested, when running as a pre-commit hook, or any time uncommitted code changes need security validation before they enter the repository. Only report findings of low severity or above. Critical, high, and medium findings block the commit; low findings warn without blocking.

Perform a comprehensive pre-commit IaC compliance review on uncommitted infrastructure-as-code changes (Terraform, CloudFormation, Bicep, Pulumi, Ansible, Kubernetes manifests, Helm charts, CDK, and similar). Checks staged changes against CMS ARS 5.1 and NIST SP 800-53 Rev 5 controls that are directly enforceable through IaC. Blocks on critical, high, and medium findings; warns on low. Use whenever an IaC compliance review is requested, when running as a pre-commit hook, or any time uncommitted infrastructure changes need compliance validation before they enter the repository. Only report findings of low severity or above.

Perform a comprehensive security and IaC-compliance audit of an entire existing codebase, directory-by-directory. Unlike the pre-commit and PR review skills which review diffs, this skill reviews the full content of files at rest. Designed for one-time baseline audits, periodic full reviews (e.g., quarterly), audits before a compliance assessment, and audits of repositories being onboarded into a regulated environment. Produces one markdown report per directory batch in audit-reports/, optionally with SARIF output for ingestion into security dashboards. Supports min-severity filtering and resume mode for long-running audits.

Independent second-opinion adjudication of an automated first-pass security or compliance review. Given a first-pass findings report, re-inspect the cited code with fresh eyes and classify each finding as confirmed, false positive, or overstated (downgraded), with a one-line rationale each. Removes false positives before they reach the developer or report, without any manual suppression list. Invoked automatically by the dispatcher when a first pass reports findings; the gate is then decided by the adjudicated result.

Applies Tailwind-only styling with no inline styles, following the Agentic Submission V2 theme from tweakcn. Use when designing UI, building or styling components, or when the user requests styling that matches the project style guide.

Applies Tailwind-only styling with no inline styles, following the Agentic Submission V2 theme from tweakcn. Use when designing UI, building or styling components, or when the user requests styling that matches the project style guide.

Principal-engineer codebase health pipeline — a deliberate pause-between-features to buy back optionality. Runs five phases: Discovery (one-concern-per-subagent fan-out over code debt, smells, architectural decay, separation of concerns, spent optionality, doc gaps, oversized units, AI-generated-code & security smells, and temporal decay — returns finding+evidence with NO severity), Verification (an independent blue-team pass that scores validity + impact and drops low-confidence / low-value findings), Remediation (blind parallel proposers with move-level convergence + an OSS-research tiebreak), Approval (a Remediation Plan reviewed one item at a time, approve/refine/reject), and Ticketization (approved work filed as child tickets under a Janitor Cleanup epic). Never edits code — the plan and tickets are the deliverable. Use when the user wants a maintainability/tech-debt review, asks to "clean up" or "audit" a codebase, or invokes /rebar-janitor.

A hypothesis-driven debugging protocol. Takes a problem (or list of problems) and drives each to a proven root cause via the scientific method under a strict two-phase discipline. Phase 1 (Understand) is analysis-only and moves through three separated stages — gather context into a descriptive dossier, fan out a subagent to propose falsifiable root causes with citations, then run experiments to confirm/reject (looping on total rejection) — and its work product is a confirmed, cited root cause. Only a proven root cause unlocks Phase 2 (Repair), which enforces strict RED→GREEN discipline: write a failing test that exercises the mechanism, fix it with a held-out subagent, and replicate the original report to confirm resolution. Use when the user wants to debug a bug/incident/ticket rigorously, asks to find the root cause of a failure, or invokes /rebar-debug.

Use this skill before your first browser action. Covers snapshot workflow, selectors, masked fields, modals, Cloudflare Turnstile, form submission protocol, and forbidden actions. Has reference files for specific situations like disabled submit buttons, custom dropdowns, and CAPTCHA handling.

Use this skill when interacting with a caseworker. Covers plain-language communication rules, gap analysis protocol (when to call gapAnalysis tool and how), form summary protocol (when to call formSummary tool and how), and step-limit handling.

Creates new Playwright end-to-end tests for the oscer reporting-app following the Page Object Model pattern. Use this skill whenever the user wants to write a new e2e test, add test coverage for a UI flow, or says things like "write an e2e test for...", "add a playwright test for...", "test the [flow name] flow", or "create a spec for...". Also trigger when the user describes a UI flow and asks for test coverage without explicitly mentioning Playwright. Always invoke this skill before writing any test code — it handles server availability checking, live app exploration via Playwright MCP, and generates idiomatic test files that match the project's existing conventions.