opentraces

Share agent traces to open datasets on HuggingFace Hub. Use this skill when the user mentions OpenTraces, trace capture, Trace Trails, workflow-built datasets, dataset review, or publishing reviewed dataset rows.

Share agent traces to open datasets on HuggingFace Hub. Use this skill when the user mentions OpenTraces, trace capture, Trace Trails, workflow-built datasets, dataset review, or publishing reviewed dataset rows.

Project already-captured traces into scored-rollout rows for the SkillOpt optimizer.

Per-commit trace lineage rows for a branch context, deterministic.

Derive compact eval rows from command-attributed trace trajectories.

Land every site-related fix on main, deploy opentraces.ai to Vercel production, and verify it works across all pages (desktop + mobile) and the email-capture flow, fixing and re-deploying until green. Runs as a goal: land -> deploy -> verify -> triage -> repeat until safe. Use when the user says "deploy site", "deploy to vercel", "push site", "ship the site", "deploy-site", or "land and verify the site". For a full coordinated release that also bumps the version and publishes the Python packages, use /release-pack instead.

Take one or more GitHub issues and drive each to a review-ready pull request, holding one belief throughout: an issue is a hypothesis, not a spec. The pipeline VERIFIES BEFORE IT TRUSTS and runs TWO ADVERSARIES before it commits: multi-agent investigation of the real cause (flagging any "landmine" — a cited function or file:line that does not exist) and the blast radius, a plan whose test must go RED BEFORE GREEN, two adversarial gates on that plan (a self/ethos checklist + a pure codex adversarial pass), implementation expressed as a forged `/goal` (via goal-forge) whose verify loop runs the otbox acceptance journey, a green suite + docs sync, a PR that `Closes #<n>`, and a FINAL codex adversarial review of the diff. Honesty, not theater: it reports unmet acceptance criteria and invalid issues rather than faking a fix. The human merges. Everything happens in a per-issue git worktree; multiple issues each get their own worktree + branch + PR. Use when the user says "issue-to-pr", "solve issue", "fix issue #N"

Synchronize all documentation surfaces after code changes. Use when the user says "docs-update", "update docs", "sync docs", "update references", "docs are stale", or after any significant code change (CLI commands, schema fields, install methods, new features). This skill catches stale references across 50+ files: marketing site, mkdocs pages, llms.txt, the agent skill, CLI help text, viewer UI, and READMEs. Proactively suggest this after merging features or changing CLI surface.

Project Trace Index skill_invocation units into reviewed skill-use episodes.

Full coordinated release of all opentraces packages — schema, CLI, the opentraces-pi npm extension, and the marketing site — in a single orchestrated flow. Use when the user says "release everything", "full release", "release pack", "release all", "cut a full release", "ship it all", "release-pack", or when doing a versioned release that should touch all packages at once. This is the right skill any time you want to go from current code to a published, verified release across PyPI (schema + CLI), npm (opentraces-pi), GitHub Releases, Homebrew, and Vercel in one pass. It handles version bumps, docs checks, ordered publishing, propagation waits, and post-release verification of pipx, brew, and npm installs automatically.

Search and use local OpenTraces agent traces, inspect capture status, and build datasets from captured Pi/Claude/Codex sessions.

Convert reviewed skill episodes into leakage-safe eval task rows.

Store synthetic otbox-style rollouts for a candidate skill version.

Emit human-gated SkillOpt update report rows for Skill Intelligence.

Author a trace-grounded, CALIBRATED verifier for one skill, two ways: an interactive ALIGNMENT SESSION (manual) or AUTOVERIFY (the agent self-aligns to the skill's goal). Use this skill when asked to build, align, calibrate, or autoverify a verifier/rubric for a skill. The agent proposes a rubric; the factory scores it against evidence + labels; a human approves promotion.

Mine latest-generation skill_invocation evidence into verifier candidate rows (one row per skill x archetype) for the Skill Verifier Factory.

Share agent traces to open datasets on HuggingFace Hub. Use this skill whenever the user mentions sharing, publishing, or uploading traces or agent activity to HuggingFace. Also use when user says "opentraces", "share this trace", "publish traces", "upload traces", "contribute traces", "donate traces", or asks about trace review, redaction, staging, or push workflows. Proactively suggest this skill after completing significant agent traces where valuable work was done.

Snapshottable full test environment for opentraces. Provisions isolated boxes (local fs, docker, or SSH-leased remote), seeds a fully-populated world, runs declarative journey TOMLs against tiered (bronze/silver/gold) checkpoints, supports artifact-preferred captures with synthetic fallback, and tears down with zero host residue.

Design and test OpenTraces local dataset workflows that emit schema-valid JSONL rows.

Build rows for the grill-me-intents dataset.

Release a new version of the opentraces CLI to PyPI, Homebrew, and GitHub. Use when the user says "release", "new release", "bump version", "publish new version", "release opentraces", "cut a release", or "release-cli". This handles the CLI package only, not the schema package separately. For a full coordinated release of schema + CLI + site together, use /release-pack instead.

Generate branded marketing screenshots of opentraces CLI commands — wordmark + CLI card + caption + wallpaper, filename-stamped and auto-copied to the clipboard. Use this skill whenever the user wants to create a shareable image, social post, promo graphic, or marketing material featuring an `ot` / `otd` / `opentraces` command, or says "ot-shot", "marketing shot", "take a screenshot of this command", "make this tweetable", "social card for X", "promo graphic", or "dress this up for sharing". Also use proactively after running an interesting opentraces command the user might want to share externally.

Create a new plan document in kb/plans/ for the opentraces project. Use this skill whenever the user wants to capture a new feature idea, operational task, bug fix, or refactor as a plan — even if they just say "let's plan X", "add a plan for Y", "write up a plan", "new plan", or describe an idea they want to backlog. The skill interviews the user to reach sufficient depth before writing, then produces a complete plan file following the project's established template and naming convention.

Sync both git repos in this project: opentraces (root) and closedtraces (kb/). Runs the full lifecycle for each — pull, check status, commit any pending changes, push. Use when the user says "gitsync", "sync repos", "sync both repos", "commit and push both", or "sync opentraces and closedtraces".

Release a new version of the opentraces-schema package to PyPI. Use when the user says "release schema", "bump schema", "schema release", "publish schema", "release-schema", or when schema models have changed and need publishing independently of the CLI. This handles the schema package only, not the CLI. For a full coordinated release of schema + CLI + site together, use /release-pack instead.