fullsend

PR review orchestrator. Triages the change, dispatches specialized sub-agents in parallel across review dimensions, synthesizes their findings, runs PR-specific checks, and produces a structured review result. Sub-agent definitions live in sub-agents/ relative to this file.

Check whether ADR numbers in the current branch conflict with ADRs already on the PR's target branch, and renumber if needed. Use before merging a PR that adds new ADRs, to avoid number collisions with concurrently merged ADRs.

Use when the user wants to tag a release, cut a release candidate, or ship a new version. Also use when asking about release process, versioning, or how GoReleaser is configured.

Standalone procedure for reviewing any code change. Identifies the change, reads surrounding source, evaluates across six review dimensions, and compiles structured findings. Can be invoked directly for local review or delegated to by the pr-review skill.

Use when checking e2e test health or reviewing recent e2e failures on main.

Use when you need to analyze the delta between agent review and human review on a PR to identify structural repo improvements that would close review gaps or justify increased agent autonomy.

Step-by-step procedure for implementing a GitHub issue. Gathers context, discovers repo conventions, plans the change, implements, verifies with tests and linters, and commits to a feature branch.

Discover repository labels and recommend contextual labels to add or remove on issues and pull requests. Produces label_actions in the agent result JSON.

Use when writing, proposing, or accepting Architecture Decision Records (ADRs) in this repo. Use when a decision has crystallized from a problem doc and needs to be recorded, or when updating living documents after an ADR is accepted.

Use when writing, editing, or adding user-facing documentation under docs/guides/. Use when creating admin guides (install, configure, manage) or user guides (workflows, interactions, interventions) for fullsend.

Use when you need to add a PR to a GitHub merge queue, check what's currently queued, or find out why a PR was removed from the queue. The gh CLI has no built-in merge-queue commands, so this skill provides scripts that use the GraphQL API.

SRE runbook for enrolling new GitHub orgs or repos into the fullsend token mint service using the fullsend CLI. Use when onboarding a new org, adding a per-repo WIF provider, or re-enrolling after infrastructure changes.

Step-by-step procedure for addressing review feedback on an existing PR. Reads review comments, plans targeted fixes, implements, verifies with tests and linters, commits, and produces structured output for the post-script.

Evaluates whether a PR's code changes have made in-repo documentation stale. Discovers documentation files, matches changed identifiers against doc content, and produces findings for stale or missing documentation. Read-only — flags staleness but does not update docs. Can be delegated to by the pr-review skill.

Use when a GitHub spec PR needs review and PR comments ingested into spec.md and qna.md: prepare comments.md from GitHub (with edit detection and qna.md placement rules), run spec-refine phases, commit spec.md and qna.md only (never comments.md), then push and reply on threads—or defer those writes to harness post-automation. Pair with spec-refine and spec-start for formats.

Use when an existing docs/plans topic from spec-start needs review feedback folded back in: optional comments.md into qna.md ### answers, resolved threads into spec.md, and the same self-review bar as spec-start. Documentation only inside the topic directory; requires spec.md and qna.md already present. Pair with spec-refine-github for PR round-trips.

Use when a GitHub issue is the canonical prompt for a spec-start run: fetch the issue, work on agent/<issue>-spec-* branches, commit planning artifacts, then open a PR locally or stop before push when a harness post-script owns GitHub writes. Pair with spec-start for the headless planning procedure.

Runs one headless pass that turns an underspecified prompt into a topic directory under docs/plans/ with spec.md and qna.md only, using the same design discipline as an interactive brainstorm but without blocking questions. qna.md uses a flat ## Q-NN log (Kind bullets + optional ### answers; one example ### under Q-01). Use for one-shot spec generation, GitHub-bound planning agents, or when the user invokes spec-start without back-and-forth clarification. For issue URL as canonical prompt plus branch/PR flow, pair with spec-start-github.

Build a merged RICE priority table: top unassigned backlog issues plus issues assigned to the current user. Use for /topissues or when the user asks for highest-priority issues, RICE scores, or backlog overview.

Use when performing a retrospective on an agent workflow. Teaches how to trace workflow runs, explore context with subagents, and write structured improvement proposals.

Analyze fullsend agent run transcripts from GitHub Actions artifacts. Use when the user wants to inspect what an agent did during a run, debug failures, check tool usage, or search transcript content. Handles downloading artifacts, finding JSONL files, and running the analyzer script.

Use when an agent hasn't posted results, a workflow run failed, or you need to find the GitHub Actions run for a fullsend triage, code, or review agent given an issue number or PR number

Use when diagnosing why agents underperform or ideating improvements to agent infrastructure — skills, agent definitions, harness configs, AGENTS.md files, hooks, CI gates, or context files.

File well-crafted GitHub issues. Use when the user wants to report a bug, request a feature, propose a change, or file any GitHub issue. Searches for duplicates, asks clarifying questions, and creates the issue using the gh CLI.

Use when the user wants to replay a Claude Code session from a GitHub Actions artifact link. Downloads the artifact, finds JSONL files, generates an interactive HTML replay with claude-replay, and serves it locally.